ArticlePDF Available

Issue Voting in Gubernatorial Elections: Abortion and Post-Webster Politics


Abstract and Figures

Nearly all studies of gubernatorial voting focus on the role of state economic conditions and incumbency on vote choice. Yet gubernatorial campaigns frequently focus on social issues such as abortion, the death penalty, and gun control. Using data from 1989 and 1990 exit polls in 10 states, we find that abortion was a significant predictor of vote choice in all but one. Our logistic regression analysis suggests that abortion position had a greater impact on vote choice than state economic conditions in eight of the 10 states in our analysis, and that abortion was a stronger predictor than even partisanship in Pennsylvania. This suggests the need to consider noneconomic issues in gubernatorial voting studies.
Content may be subject to copyright.
Issue Voting in Gubernatorial Elections: Abortion and Post-Webster Politics
Author(s): Elizabeth Adell Cook, Ted G. Jelen and Clyde Wilcox
The Journal of Politics,
Vol. 56, No. 1 (Feb., 1994), pp. 187-199
Published by: The University of Chicago Press on behalf of the Southern Political Science
Stable URL:
Accessed: 27-07-2017 18:08 UTC
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted
digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about
JSTOR, please contact
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
Southern Political Science Association, The University of Chicago Press
are collaborating
with JSTOR to digitize, preserve and extend access to
The Journal of Politics
This content downloaded from on Thu, 27 Jul 2017 18:08:21 UTC
All use subject to
Issue Voting in Gubernatorial Elections:
Abortion and Post-Webster Politics
Elizabeth Adell Cook
The American University
Ted G. Jelen
Illinois Benedictine College
Clyde Wilcox
Georgetown University
Nearly all studies of gubernatorial voting focus on the role of state economic conditions and in-
cumbency on vote choice. Yet gubernatorial campaigns frequently focus on social issues such as abor-
tion, the death penalty, and gun control. Using data from 1989 and 1990 exit polls in 10 states, we
find that abortion was a significant predictor of vote choice in all but one. Our logistic regression
analysis suggests that abortion position had a greater impact on vote choice than state economic con-
ditions in eight of the 10 states in our analysis, and that abortion was a stronger predictor than even
partisanship in Pennsylvania. This suggests the need to consider noneconomic issues in gubernatorial
voting studies.
M ost studies of gubernatorial elections have used aggregate analysis to in-
vestigate the role of economic issues and incumbency on candidate success.
Chubb (1988) reported that institutionalization has partially insulated incumbent
governors from state economic conditions, but that state and especially na-
tional economic conditions still affect the outcome of gubernatorial elections.
Holbrook-Provow (1987) also found that national economic conditions are im-
portant predictors of electoral outcomes, and added that presidential popu-
larity and incumbency are also significant factors. The latter finding supports
Piereson's (1977) report that incumbency has become a more important predictor
of gubernatorial voting while party has declined in significance.
Although these and other studies focus on the role of the economy in guberna-
torial elections, there has been much less work on the impact of noneconomic is-
sues. Indeed, Chubb argued that the "unpredictable" portion of the variation in
gubernatorial outcomes must be due to candidate characteristics, not noneconomic
issues. We argue, however, that noneconomic issues can play a major role in state
The data were made available by the Inter-University Consortium for Political and Social Research.
We alone are responsible for all interpretations.
THEJOURNAL OF POLITICS, Vol. 56, No. 1, February 1994, Pp. 187-99
C 1994 by the University of Texas Press, P.O. Box 7819, Austin, TX 78713-7819
This content downloaded from on Thu, 27 Jul 2017 18:08:21 UTC
All use subject to
188 Elizabeth Adell Cook, Ted G. Jelen, Clyde Wilcox
gubernatorial elections. Gubernatorial candidates frequently stress noneconomic
issues in their campaigns. In 1988 and 1990, candidates touted their positions on
education, environment, death penalty, gun control, and abortion.
Moreover, although governors can have some influence on the economic for-
tunes of their state, national economic conditions have a profound impact on state
economies. Voters may be aware that governors are more likely to have an impor-
tant impact on the other issues listed here. For some voters, such issues may have
a bigger influence on their gubernatorial vote than state economic fortunes.
The Webster v. Reproductive Health Services decision in 1989 seemed to invite
state governments to regulate access to abortion. Although prochoice voters had
previously believed that the abortion right created in Roe v. Wade was immune
to interference by state legislatures and governors, in the aftermath of Webster
a number of states moved to restrict legal abortion. Prochoice citizens became
aware of the political consequences of electing prolife officials, and prolife citi-
zens saw the opportunity to enact at least some restrictions on legal abortion in
many states.
Gubernatorial candidates in 1989 and 1990 stressed the abortion issue in their
campaigns (Donovon 1990; Idelson 1990). Some scholarly studies have shown that
the abortion issue was a significant factor in the outcome of governors elections in
1989 in Virginia and New Jersey (Dodson and Burnbauer 1990; Cook, Jelen, and
Wilcox 1991), and in Louisiana in 1991 (Howell and Sims 1993). Other research
suggests that abortion may have played a role in electoral outcomes in 1990 as well
(Cook, Jelen, and Wilcox 1992).
Previous research may have ignored the role of noneconomic issues because it is
difficult to incorporate such variables into aggregate models. Data on state eco-
nomic conditions are relatively easy to obtain, and in longitudinal analysis in-
formation on incumbency and presidential popularity is readily available. Yet to
study the impact of noneconomic issues requires individual data from random
samples of state voters. Until recently, such data were rare. In this paper, we ex-
amine the role of the abortion issue in gubernatorial elections in 1989 and 1990.
Using data from exit polls, we focus on elections in a number of key states that
produced prochoice and prolife victors for each party.
In 1989, in the immediate aftermath of Webster, Virginia and New Jersey elected
prochoice Democratic governors over prolife Republican candidates.' In 1990,
prochoice Democrats beat Republicans who took prolife stands in Texas and
Florida. Prochoice Republicans were elected in California, Massachusetts, and
IFor this paper, we use the term prochoice to describe the views of candidates who favor gener-
ally unlimited access to abortion, and prolife to describe candidates who favor sharp restrictions on
legal abortion. Some prolife activists might object to the use of this label for a candidate who would
allow abortion under a few limited circumstances, but we have adopted the term because of its
widespread use in the popular media and because candidates who take a strictly prolife position are
very rare.
This content downloaded from on Thu, 27 Jul 2017 18:08:21 UTC
All use subject to
Issue Voting in Gubernatorial Elections 189
Illinois. In Illinois, the Democratic candidate had advocated restrictions on abor-
tion, and in Massachusetts the Republican candidate took a stronger position in
support of abortion rights,2 but in California the Democratic candidate argued
that her gender made her even more committed to abortion rights. Republicans
who favored restrictions on abortion were elected in Ohio and Michigan; the Ohio
contest featured a formerly prolife Democrat who had recently changed his posi-
tion. A largely prolife Democrat was elected in Pennsylvania over a prochoice
Republican. Thus these states provide examples of victory by Democrats and
Republicans who took prochoice and prolife stands.
The Data
The data come from CBS News/New York Times exit polls in Virginia and
New Jersey in 1989, and the Voter Research and Surveys General Election Exit
Poll in 42 states in 1990. The 1989 CBS surveys included this abortion item:
Which of these comes closest to your view?
1. Abortion should be generally available to those who want it.
2. Abortion should be available, but under more strict limits than now.
3. Abortion should be strictly prohibited.
The 1990 exit polls contained a somewhat different question:
Which of these statements comes closest to your view about abortion?
1. It should be legal under all circumstances.
2. It should be legal only in some circumstances.
3. It should not be legal in any circumstances.
In addition, each survey contained a measure of general ideology and of partisan-
ship. The Virginia and New Jersey surveys also included an item that tapped
previous vote history and approval of the outgoing governor. In Virginia, the out-
going governor was a Democrat, while in New Jersey he was a Republican. The
1990 surveys contained items that measured the respondents' evaluation of the
condition of the state economy, the national economy, and of her or his personal
All surveys contained measures of race, sex, age, education, income, and reli-
gious denomination. We have created two dummy variables from this latter item
to identify Catholics and those with no religious affiliation. Unfortunately, we are
unable to distinguish between evangelical and mainline Protestants with these
data. In the multivariate analysis, the effects of these variables are assessed relative
to Protestants. In all states except New Jersey the surveys also included an item
that tapped frequency of church attendance, and in all surveys except Virginia an
item identified those in union households. Virginia voters were asked whether
they grew up in the South or the North.
2William Weld ran a strong prochoice campaign during his primary election campaign in which he
defeated a prolife Republican for the nomination.
This content downloaded from on Thu, 27 Jul 2017 18:08:21 UTC
All use subject to
190 Elizabeth Adell Cook, Ted G. Jelen, Clyde Wilcox
Although it is possible to have a somewhat more narrow list of independent
variables common across all surveys, we have elected to include every relevant in-
dependent variable included in the survey in a particular state in our multivariate
analysis. We have also estimated equations that include only the common inde-
pendent variables, and the results are not substantively different.
Abortion Attitudes and Vote Choice
In table 1 we present the distribution of attitudes toward abortion, and the
salience of abortion in each state electorate. Recall that the wording of the abor-
tion item differed somewhat in Virginia and New Jersey compared to the other
eight states. The wording in those states most likely resulted in somewhat more
prochoice respondents than the 1990 item would have identified.3 In addition,
however, the Virginia and New Jersey surveys were held immediately after the
Webster decision, and featured strong efforts by prochoice groups. In Virginia,
for example, NARAL ran independently funded advertisements on behalf of
the Democratic candidate, and the Democratic candidate for Lt. Governor also
stressed the abortion issue in his advertising (Cook, Hartwig, and Wilcox 1992).
For all of these reasons, it is important to compare the attitudes of Virginia and
New Jersey voters only to those in the other state and not in the eight states sur-
veyed in 1990.
In each state, there are more prochoice than prolife respondents, but in most
states a plurality of respondents favor some restrictions on abortion. In Virginia
and New Jersey in 1989, an absolute majority favored legal abortion. In California
in 1990, a majority of respondents favored keeping abortion legal, while in Texas,
Ohio, Pennsylvania, and Illinois the prochoice contingent comprised fewer than
40% of the electorate.4
Respondents were asked which issues were the most important in deciding their
vote and which issues were the second-most important. In table 1, we see that abor-
3The 1990 surveys included a sample in New Jersey, where 40% took a consistent prochoice posi-
tion, 44% a middle, or situational position, and 16% a prolife position. Elsewhere we have shown that
respondents to general abortion items tend to select the closest possible category. If the middle cate-
gory is relatively liberal, those who favor abortion under only a very few circumstances (e.g., woman's
health and rape) will choose a prolife response rather than agree to a far more liberal response; if the
middle category is relatively conservative then those who approve of abortion in most but not all cir-
cumstances will choose a prochoice position. The 1989 item mentioned "strict" restrictions, while the
1990 question merely noted that it should be legal in some circumstances. Respondents may have per-
ceived that the middle option in the 1989 item was somewhat more conservative. In fact, general items
usually overestimate support for a prochoice and a prolife position. Of course, the difference in re-
sponses between 1989 and 1990 may also have been due to the different context. The 1989 survey was
conducted immediately after Webster and after a campaign that focused on abortion.
4General three-point items such as this overstates the support for both the prochoice and prolife
position. In addition, many of those who take prochoice positions on these types of items favor some
restrictions such as parental notification or consent.
This content downloaded from on Thu, 27 Jul 2017 18:08:21 UTC
All use subject to
Issue Voting in Gubernatorial Elections 191
Importance of Abortion
1st 27% 30% 10% 12% 12% 7% NA 23% 16% 16%
2nd 23% 3% 17% 15% 17% 7% NA 15% 13% 15%
Abortion Should Be:
Legal 50% 57% 47% 33% 54% 47% 37% 43% 36% 31%
Limited 33% 28% 41% 52% 34% 40% 47% 44% 49% 50%
Not Legal 18% 16% 12% 16% 12% 14% 16% 14% 16% 20%
Percentage of respondents in each state who listed abortion as the most important or second most
important issue in their vote decision, and who took each position on the legality of abortion.
tion was highly salient to voters in 1989 and 1990. Fully half of Virginia voters cited
the abortion issue as important in their vote, as did nearly 40% of voters in
Michigan and nearly one-third of voters in Pennsylvania. Only in Massachusetts
was the issue relatively unimportant to the electorate.
In table 2, we show the bivariate relationship between abortion attitudes and
vote choice, and provide comparable information on partisanship. Although the
relationship between abortion attitude and vote is not surprisingly weaker than
the effects of partisanship, abortion appears to be a source of vote choice in each
state except Michigan. Note that abortion attitudes matter in states such as
Pennsylvania, Illinois, and Massachusetts, where the candidates took positions
that deviated from those of the national parties. The voters seem to have sorted
out these positions, for prolife voters were more likely in those states to support
the Democratic candidate and prochoice voters the Republican candidate.
Also in table 2 we compare the effects of abortion attitudes, holding constant
partisanship. These data show that the abortion issue influences vote choice
among partisans but has a somewhat stronger impact among independents. Note
the strong impact of the abortion issue among voters in Virginia and Pennsylvania
after controls for partisanship.
Finally, in table 2 we show the percentage of all voters in each state who men-
tioned abortion as a salient issue and who took a prochoice or prolife position.
This is one measure of the size of the prolife and prochoice voting blocs in each
state. The figure is a result of the size and intensity of the prolife and prochoice
'Of course, many of those who favored abortion in some but not all circumstances also mentioned
abortion. These situational voters may lean toward one candidate or another depending on the posi-
tions of the candidates and the frame of the issue.
This content downloaded from on Thu, 27 Jul 2017 18:08:21 UTC
All use subject to
192 Elizabeth Adell Cook, Ted G. Jelen, Clyde Wilcox
Percentage Voting for Democratic Candidate
Abortion Should Be:
Legal 70% 70% 69% 67% 58% 44% 53% 56% 58% 45%
Limited 37% 56% 46% 48% 40% 50% 42% 58% 40% 73%
Not Legal 18% 48% 43% 36% 37% 59% 53% 50% 29% 88%
Democrats 85% 88% 87% 84% 83% 67% 76% 79% 76% 80%
Independents 50% 69% 59% 49% 51% 44% 46% 43% 38% 66%
Republicans 18% 27% 26% 19% 13% 24% 16% 22% 15% 50%
Legal 92% 90% 95% 91% 87% 60% 78% 80% 84% 62%
Limited 81% 87% 79% 82% 78% 71% 71% 79% 70% 85%
Not Legal 51% 84% 79% 73% 74% 78% 85% 76% 74% 92%
Legal 68% 76% 67% 68% 62% 39% 51% 42% 53% 41%
Limited 38% 63% 52% 40% 40% 48% 41% 45% 33% 75%
Not Legal 21% 48% 40% 23% 29% 48% 45% 37% 17% 97%
Legal 35% 35% 40% 31% 17% 21% 20% 17% 20% 27%
Limited 13% 22% 18% 15% 11% 21% 14% 27% 15% 51%
Not Legal 2% 13% 16% 12% 4% 42% 16% 23% 5% 77%
% of Voters Who
Mention Abortion
as a Key Issue
and Want Abortion:
Legal 21% 23% 5% 8% 13% 7% NA 15% 9% 9%
Not Legal 12% 15% 1% 5% 4% 2% NA 4% 6% 7%
N 1,085 1,559 1,625 2,404 2,766 1,931 2,864 1,565 1,666 1,539
Note: In Virginia, New Jersey, Florida, Texas, California, Michigan, and Ohio, the Democratic can-
didates took prochoice stands. In California, Massachusetts, Illinois, and Pennsylvania, the Republican
candidate took a prochoice position.
blocs. The prochoice forces enjoy a numerical advantage in all states, but the prolife
contingent viewed abortion as more salient in all states. Nonetheless, in all 10 states
there were more prochoice voters who mentioned the abortion issue as salient
than prolife voters. The prochoice advantage was large in Virginia, California, and
Michigan, and small in Texas, Ohio, and Pennsylvania.6
6It should be noted that a number of voters who favored allowing abortion under some but not all
circumstances also listed abortion as a salient issue.
This content downloaded from on Thu, 27 Jul 2017 18:08:21 UTC
All use subject to
Issue Voting in Gubernatorial Elections 193
Multivariate Results
To see if abortion attitudes predict vote choice after controls for other variables,
we have estimated separate logistic regression equations for each state. In table 3,
we show the results for Virginia and New Jersey in 1989. The coefficients are un-
standardized and cannot be directly compared within equations to determine the
relative impact of each variable on vote choice. The -2 Log LR (log likelihood) is
Virginia New Jersey
MLE -2 Log LR MLE -2 Log LR
Partisanship 1.03*** 43.81 1.44*** 192.10
Party Vote History 1.05*** 34.07 .75*** 22.17
Ideology .60*** 11.72 .73*** 36.76
Approval of Incumbent
Governor .72*** 7.32 -1.32*** 50.56
Race 2.79*** 42.63 .38 1.66
Sex .27 0.90 .03 .01
Age .22* 2.80 -.05 .45
Education -.19* 3.15 -.02 .06
Income -.01 0.19 .05 .59
Raised in South 1.01*** 14.82
Union Household -.04 .04
Catholic .15 .19 .07 .00
No Denomination -.42 .54 -.71* 3.17
Church Attendance -.09 .80
Abortion position 1.06*** 37.13 .62*** 30.14
Prochoice .95*** 18.22 .60*** 10.75
Prolife -1.21*** 10.94 -.63*** 6.18
Percentage Predicted Correctly:
Democrat 84% 86%
Republican 87% 72%
Overall 86% 81%
Modal value 62% 50%
Goodman & Kruskal's Tau .37 .28
Model Chi-Square 539*** 545***
Unweighted N 823 1,130
up ? .10; **p ? .05; ***p ? .01. Coefficients for prochoice and prolife dummy variables are from a
separate model, which included the two dummy variables instead of the single abortion item.
This content downloaded from on Thu, 27 Jul 2017 18:08:21 UTC
All use subject to
194 Elizabeth Adell Cook, Ted G. Jelen, Clyde Wilcox
similar to a T-value in regression and is preferable to the Wald test for the sta-
tistical significance of the coefficients (Hosmer and Lemeshow 1989; Hauck and
Donner 1977; Jennings 1986).7
To assess the substantive importance of abortion, we have estimated separate
equations in which we entered abortion attitudes as two dummy variables identi-
fying those who took a prochoice and prolife position, leaving those who sup-
ported legal abortion under some but not all circumstances as the excluded
category.8 These coefficients are denoted by two asterisks and appear below the
coefficient for the single abortion item. We have converted these coefficients into
probabilities, which we will discuss later.9
In both Virginia and New Jersey, abortion is a significant predictor of guberna-
torial vote. Note that we control for both partisanship and past party vote, so that
the effects of party loyalty is shared between these two variables. Abortion was a
strong predictor of vote choice in each state, and the analysis with the two dummy
variables shows that both prolife and prochoice citizens were likely to vote their
abortion preferences. In New Jersey, after holding constant demographic charac-
teristics, ideology and partisanship, prolife voters had a .47 probability of voting
for the Democratic candidate compared with a .61 probability for prochoice vot-
ers. In Virginia, prolife voters had a .39 probability of casting a Democratic ballot
while prochoice respondents had a .64 probability of doing so.
In table 4, we show the equations for the 1990 gubernatorial contests. Abortion is
a statistically significant predictor in each of the states except Michigan and is a
strong predictor in six of the eight states. We experimented with other specifica-
tions, including removing the item on the respondent's personal finances to include
only evaluations of the state economy, substituting the respondents evaluations of
the national economy in those states in which the item was asked, and including all
three measures of economic attitudes. The results were substantially identical to
those presented here: abortion attitudes remain a strong, statistically significant
predictor of vote choice. In Pennsylvania, abortion attitudes were especially power-
ful predictors of vote: after multivariate controls the probability of a Democratic
vote by a prolife citizen was .81 compared with only .42 for a prochoice voter.
Because the logistic regression coefficients are unstandardized, it is difficult to
compare the impact of variables within an equation. Moreover, there is no widely
7We used the SPSS-X mainframe version of logistic regression.
8See Howell and Sims (1993) for a discussion of this procedure.
9Logistic regression coefficients provide an estimate of the log odds after adjusting for other in-
dependent variables. The log odds is the log of the odds ratio for x = 1 (e.g., prochoice) and x = 0
(e.g., not prochoice). The log odds can be transformed into probabilities which assume that the other
independent variables in the model assume the value of the sample mean. The procedure involves
computing a variable x for each case that represents a linear combination of the unstandardized
logistic regression coefficients. The calculation involves raising e to the X power, and dividing by
1 + e to the X.
This content downloaded from on Thu, 27 Jul 2017 18:08:21 UTC
All use subject to
Issue Voting in Gubernatorial Elections 195
accepted procedure for producing standardized coefficients. We transformed the
coefficients for partisanship and for evaluations of the state economic conditions
into probabilities (now shown), and compared the impact of these variables on
vote choice. In almost all states, the differences in probabilities of Democratic
voting between Democrats and Republicans were far greater than those between
prolife and prochoice voters, although in Pennsylvania abortion appears to have
swayed slightly more votes.'0 In eight states the probabilities associated with
abortion suggested a greater impact on vote choice than evaluations of the state
Not surprisingly, abortion was an especially strong predictor in those states in
which at least one candidate chose to focus on the issue. In Florida, the Re-
publican incumbent had called a special legislative session in an unsuccessful ef-
fort to limit abortion, although the Democratic candidate did not emphasize the
issue. Yet prochoice voters responded negatively to the position of the Repub-
lican incumbent while prolife voters were not significantly different from those
who took a moderate position on abortion. In Pennsylvania, the incumbent Demo-
cratic governor had signed a restrictive measure in 1989, and his prochoice Re-
publican opponent tried to make abortion a central issue in her campaign. In
Ohio, the Democratic candidate had changed from a prolife to a prochoice posi-
tion before the election, and the Republican candidate sought to capitalize on
that change.
Dodson and Birnbauer (1991) reported that abortion was more salient for
women than men in Virginia and New Jersey in 1989. We estimated equations
with interactions between sex and abortion attitudes for each state. In only one
state, Michigan, was there a significant interaction with women more likely to vote
the abortion issue. We also tested for the interaction between religion and abortion
positions and found significant interactions in Illinois, Pennsylvania, Texas, and
New Jersey. Catholics were more likely to cast a vote based on abortion than either
Protestants or those without religious attachments.'2
10The probability of a Republican voting for the Democratic incumbent was 49%, and the
Democratic odds were 73%.
"For example, the odds of voting for the Democratic candidate in Ohio by about 10% be-
tween those who thought the state economy was excellent or poor, but the spread on the abortion
issue was 18%. The spread of probabilities across party was 43%. Interestingly, our analysis sug-
gests that the abortion issue helped the Democratic candidate, although a good deal of attention
was paid to his recent conversion to pro-choice politics. We estimate that the abortion issue
produced about a 1.5% vote swing to the Democratic candidate, after all other variables were
12With one significant interaction in 10 equations for gender, we are not especially confident that the
result is not due to chance. At the .05 level of significance, we have a 50% chance of getting a false pos-
itive in 10 equations. The interactions for religion are more persuasive, and they are found in states
where there is some religious pluralism, and in generally in states where the Catholic church has been
active on this issue.
This content downloaded from on Thu, 27 Jul 2017 18:08:21 UTC
All use subject to
o- - c o
cn~~~~~~~~~~~~~~~~~~~~~~~~~~~~P I LnE C1 4=
u n en 0 en 00 4= 00 r- C > en - en
Ln C< II V- e o e e o - n
| ..... ....... 1~~~~~~~~~~~~~~~~~~~~~~- 0- 0-0- -, 1 --- -.....1
; N e _ ? e + ? m > X N oo + m N ? m~~~~~~~t~ C4 e el
q el I n e Nr , " 0e
I~~~~~~~~~~~~~~~~~~~e CP - C g*tN
z _. r- _r o C, (1 _0 _ ooo N ON 0
z <= -- ell
rell O + _ X ? m e X ? _ t
<~~~~~~~~~~~~~~: X > N>
i~~~~~~0 CP X tr r- r * * * **_ b
o * ** **** *** * * *CD _ _
;~~~~~~~ ~~~~ K CH C h-Xe0 -
ffi A S _ . m m O. N 1. ? _ e O. e O N ? ? ?~~~~~b
Z ; ? o t_ sD O e _ m 00 N N - ? X N o X~~~~~~~~~~~~U b
bD e ffi O. O e N + 00 o N e + 00 e > ? N~~~~~~~~~~I. 7
O~~~~~~0 z 'U Nm
This content downloaded from on Thu, 27 Jul 2017 18:08:21 UTC
All use subject to
00 oo _ l 00N N0 _ 0 0 ) eq . O0 oO L0.
O0 N0 % ' 0o 0 % ON M 00 N
; B~~~~~~~~~~~r
i~ cW
S Ch + L f N e ? o e O. O No N N fq
* *EE *X * *
*; O * * } *+ * ** ,-.o -m
'00 - NO ~~00 r-~,
D~~~e r- M .NO No C) o.~ m eq _ o _ qm
) Ln \ 0 0n "o 0 ON O" LP r- 0 oN oo ^t
N- O 00- 0N en 'I ON en 0 " 0 \ __
| , ri C'' 'i 0> '6 S pi \ r g g6#
I ;00 tN- \ * 00 00
,%; * ~) o~ X o~ n U N * *
- - * I . en._a oo oo .. oo en oo
, * * ** * ** * ** * X N X X N N o
NO 0 0 N-N N o 0' 4 g -0%
~~- N- 0-0 0 ~~~~~~~~~~~4O~~~~ e
o N0 00o |00n 000e N 0 0
0 n m1f eneq\ 000en00 -00% a N Ne' 4- C=00
.-~~~~~~~~~~~~~~~~~~~~~~~~I N N - --0-
eo'4NNO N= r 00 .
00 00 00Lf
;M \ 0 cQ 02;l< , Ut 2 E0 3 vi
- - . l~~~~l* *I* ~ ~
Co co~~~~~~~~~~~~~~~~~~~~~~~~~4
z U= V'l~~~~~
This content downloaded from on Thu, 27 Jul 2017 18:08:21 UTC
All use subject to
198 Elizabeth Adell Cook, Ted G. Jelen, Clyde Wilcox
We draw three broad conclusions from this study. First, abortion has become
an important issue in state elections. In the aftermath of Webster, voters consid-
ered the positions of gubernatorial candidates on abortion and were willing to
cross party lines to support the candidates who supported their positions. Should
the Court reverse Roe in the near future, it is likely that abortion will become an
even stronger predictor of gubernatorial vote choice. As state legislatures in many
states move to restrict abortion rights, the issue will prove significant in other state
elections as well.
Second, studies of gubernatorial elections would benefit from a consideration of
noneconomic issues. In six of the eight states in the 1990 study and both states in
1989, abortion attitudes were more important than attitudes toward the state (and
national) economy. Of course, the recession was still young in 1990, and we do not
suggest that abortion or any other social issue would be more salient than eco-
nomic conditions in the depths of an economic downturn. Yet voters may be
aware that governors can do more concretely to affect legal abortion than they can
do to improve the state or especially the national economy. Other noneconomic is-
sues, such as gun control and capital punishment, may also be important predic-
tors of gubernatorial votes.
Finally, although aggregate studies remain the only way to study a number of
governors' races in longitudinal perspective, the recent availability of state exit
polls by national media provide the possibility to examine individual vote choice
in a number of gubernatorial elections. These data may alter our understanding of
the determinants of such elections. For example, studies of congressional voting
have shown a wide disjuncture between the impact of economic variables at the
aggregate and individual level, an anomaly that has been explained by sociotropic
voting (Kinder and Kiewiet 1981) and by strategic politicians (Jacobson and
Kernell 1983). Our results hint at a similar disjuncture in gubernatorial elections,
although we find no evidence for sociotropic voting. It may well be that the im-
portance of economic issues is more evident in the results of aggregate analyses
than in the study of individual-level data. More research is needed to resolve the
Manuscript submitted 4 April 1992
Final manuscript received 3 June 1993
Chubb, John. 1988. "Institutions, the Economy, and the Dynamics of State Elections." American Po-
litical Science Review 82:133-54.
Cook, Elizabeth Adell, Ted G. Jelen, and Clyde Wilcox. 1991. "The Electoral Consequences of Abor-
tion." Presented at the annual meeting of the American Political Science Association, Washington, DC.
This content downloaded from on Thu, 27 Jul 2017 18:08:21 UTC
All use subject to
Issue Voting in Gubernatorial Elections 199
Cook, Elizabeth Adell, Frederick Hartwig, and Clyde Wilcox. 1992. "The Abortion Issue Down
Ticket: The Virginia Lieutenant Governor's Race of 1989." Women IS Politics.
Cook, Elizabeth Adell, Ted G. Jelen, and Clyde Wilcox. 1992. Between Two Absolutes: American Public
Opinion and the Politics ofAbortion. Boulder, CO: Westview Press.
Dodson, Debra, and Lauren Burnbauer. 1990. Election 1989: The Abortion Issue in New Jersey and
Virginia. New Brunswick, NJ: Eagleton Institute of Politics.
Donovan, Beth. 1990. "Taxes, Abortion Roil Waters in Gubernatorial Races." Congressional Quarterly
Weekly Report, April 21.
Hauck, Walter W., and Allan Donner. 1977. "Wald's Test as Applied to Hypotheses in Logit
Analysis." Journal of the American StatisticalAssociation 72:851-53.
Holbrook-Provow, Thomas. 1987. "National Factors in Gubernatorial Elections." American Politics
Quarterly 15:471-83.
Hosmer, David W., and Stanley Lemeshow. 1986. Applied Logistic Regression. New York: Wiley.
Howell, Susan E., and Robert T. Sims. 1993. "Abortion Attitudes and the Louisiana Governor's
Election." American Politics Quarterly. Forthcoming.
Idelson, Holly. 1990. "Budgets, Jobs and Abortion are Big Issues in States." Congressional Quarterly
Weekly Report, September 8.
Jacobson, Gary, and Samuel Kernell. 1983. Strategy and Choice in Congressional Elections. 2d ed. New
Haven: Yale.
Jennings, Dennis E. 1986. "Judging Inference Adequacy in Logistic Regression." Journal oftheAmerican
StatisticalAssociation 81:987-90.
Kinder, Donald, and D. Roderick Kiewiet. 1981. "Sociotropic Politics: The American Case." British
Journal of Political Science 11:129-61 .
Piereson, James. 1977. "Sources of Candidate Success in Gubernatorial Elections, 1910-1970." Journal
of Politics 39:937-58.
Tompkins, Mark. 1988. "Have Gubernatorial Elections Become More Distinctive Contest?" Journal of
Politics 50:192-205.
Elizabeth Adell Cook is visiting assistant professor of government, The Ameri-
can University, Washington, DC 20016.
Ted G. Jelen is professor of political science, Illinois Benedictine College, Lisle,
IL 60532.
Clyde Wilcox is associate professor of government, Georgetown University,
Washington, DC 20057.
This content downloaded from on Thu, 27 Jul 2017 18:08:21 UTC
All use subject to
... However, since 1984 there has been an increasing gap between the political parties (Jelen and Wilcox 2003) and considering attitudes influence partisanship, this may potentially increase the gap in both directions (Killian and Wilcox 2008). Previous studies further support that abortion attitudes are strongly associated with their voting tendencies for Presidential, House, and Senate elections (Jelen and Wilcox 2003;Abramowitz 1995;Smith 1994;Cook, Jelen, and Wilcox 1994a;Cook et al. 1994b;Abraham and Saunders 2008). Moreover, these attitudes are also strong enough to lead people to switch political parties (Killian and Wilcox 2008). ...
... Those who hold more conservative ideologies are more likely to denounce abortion even if it was as a result of rape than more liberal respondents. Further supporting previous studies, we argue that this finding supports the idea for a broader ideological shift (Evans 2003;Jelen and Wilcox 2003;Abramowitz 1995;Smith 1994;Cook, Jelen, and Wilcox 1994a;Cook et al. 1994b;Abraham and Saunders 2008). Results suggest that even in cases of abortions due to rape, there is a notable political and religious rift. ...
... Although some data suggest that both Democrats and Republicans are becoming more supportive of women's reproductive rights (Jelen, 2017), abortion remains a cornerstone issue in the culture war (see Koleva et al., 2012;Oldmixon, 2002). In fact, abortion attitudes can be more impactful than economic conditions on vote choice (Cook et al., 1994) and contribute to partisan sorting in which pro-life Democrats and pro-choice Republicans switch to parties that match their views on abortion (Carsey & Layman, 2006;Killian & Wilcox, 2008). Yet few studies have examined partisan differences in abortion support in multiparty systems where abortion support may be less consistently associated with left-wing political parties. ...
Full-text available
Despite being a defining issue in the culture war, the political psychology of abortion attitudes remains poorly understood. We address this oversight by reviewing existing literature and integrating new analyses of several large‐scale, cross‐sectional, and longitudinal datasets to identify the demographic and ideological correlates of abortion attitudes. Our review and new analyses indicate that abortion support is increasing modestly over time in both the United States and New Zealand. We also find that a plurality of respondents (43.8%) in the United States are consistently “pro‐choice,” whereas 14.8% are consistently “pro‐life,” across various elective and traumatic abortion scenarios. We then show that age, religiosity, and conservatism correlate negatively, whereas Openness to Experience correlates positively, with abortion support. New analyses of heterosexual couples further reveal that women's and men's religiosity decrease their romantic partner's abortion support. Noting inconsistent gender differences in attitudes toward abortion, we then discuss the impact of traditional gender‐role attitudes and sexism on abortion attitudes and conclude that, rather than misogyny, benevolent sexism—the belief that women should be cherished and protected—best explains opposition to abortion. Our review thus provides a comprehensive overview of the demographic and ideological variables that underly abortion attitudes and, hence, the broader culture war.
... First, when citizens favor a politician, they also may believe the politician to be highly credible and conclude that their favored politicians' stances on issues are right (Ziegler & Diehl, 2003). Second, citizens often favor a politician based on agreement regarding issues of importance (Campbell, Converse, Miller, & Stokes, 1960;Cook, Jelen, & Wilcox 1994;Ottati, 1990); that is, when citizens believe that a politician is right on what they consider to be key political issues, they may think that the politician is also right about other issues. Another mechanism is Heider's (1958) balance theory and its applications to political attitudes (Brent & Granberg, 1982;Kinder, 1978;Ottati, Fishbein, & Middlestadt, 1988). ...
This study investigated two information-processing modes for political messages from favored politicians: "biased" systematic processing and heuristic processing. In an experiment, college students (N = 183) with different levels of political interest received messages about unfamiliar political issues from either a favored or a less favored candidate in the 2008 U.S. presidential election. For those with low levels of political interest, source favorability had a direct effect on attitudes, indicating heuristic processing. For those with high political interest, source favorability had an indirect effect on attitudes through message-relevant thoughts, indicating biased systematic processing. Theoretical and practical implications of the findings are discussed.
... We use the Cook Partisan Voting Index ( to account for the ideological leanings of the region, at the level of the state. Used extensively in prior research [74,75], this variable provides a continuous indicator of the degree of partisanship (i.e. Democrat vs. Republican) in each state bounded between-40 (extreme liberalism) and +40 (extreme conservativism). ...
Full-text available
We investigate how mass shootings influence the stock price of firearms manufacturers. While it is well known that mass shootings lead to increased firearms sales, the response from financial markets is unclear. On one hand, given the observed short-term increase in demand, firearm stock prices may rise due to the unexpected financial windfall for the firm. On the other, mass shootings may result in calls for regulation of the industry, leading to divestment of firearms stocks in spite of short-term demand. We examine this tension using a market movement event study in the wake of 93 mass shootings in the U.S. between 2009 and 2013. Findings show that stock prices of firearm manufacturers decline after shootings; each event reducing prices between 22.4 and 49.5 basis points, per day. These losses are exacerbated by the presence of a handgun and the number of victims killed, but not affected by the presence of children or location of the event. Finally, we find that these effects are most prevalent in the period 2009–2010 but disappear in later events, indicating that markets appear to have accepted mass shootings as the “new normal.”
... The pro-choice or prolife attitudes of the winners in these contests reflect the overall attitudes of the voters. Their findings lead the authors to conclude that "studies of gubernatorial elections would benefit from a consideration of noneconomic issues" (Cook 1994). ...
Does pledge fulfilment bear any electoral consequences for government parties? While previous research on retrospective voting has largely focused on electoral accountability with respect to the economy, the theoretical framework presented in this study links government parties’ performance to their previous electoral pledges. It is argued that government parties are more likely to be rewarded by voters when they have fulfilled more pledges during the legislative term. Good pledge performance of a party is associated with the ability to maximise policy benefits (accomplishment) and to be a responsible actor that will stick to its promises in the future as well (competence). Analysing data from 69 elections in 14 countries shows that a government party's electoral outcome is affected by its previous pledge performance. A government party that fulfils a higher share of election pledges is more likely to prevent electoral losses. This finding indicates that voters react at the polls to party pledge fulfilment, which highlights the crucial role of promissory representation in democratic regimes. Surprisingly and in contrast with economic voting, there is no evidence that retrospective pledge voting is moderated by clarity of responsibility. This article is protected by copyright. All rights reserved
This study examined how presidential candidates used partisan issue linkages to discuss their abortion views over the 2008, 2012, and 2016 elections. It qualitatively examined 64 speeches, town halls, and interviews in which candidates spoke about abortion to identify trends in their rhetoric. It also measured the frequencies with which candidates used partisan messages, specific vocabulary, and issue linkages. As candidates employed stronger and more partisan issue linkages across these three elections, they transformed abortion from a stand-alone issue to one entrenched in a partisan policy package. The development of Planned Parenthood as a symbol for pro-choice positions in 2012 enabled candidates to make different and more partisan issue linkages. This study further identified candidates’ changing strategies for discussing abortion, including differences along party lines and over time. These findings carry implications for politicians, voters, and scholars alike. They suggest that the abortion debate is dynamic and deserving of ongoing research. Future studies on partisan rhetoric should account for issue linkages to more accurately examine trends in partisanship.
They have money, influence, power - and they turn out to vote. “They” are groups like Focus on the Family, Family Research Council, and Concerned Women for America (all parts of the Christian Right. But, are they a serious threat to religious liberty, bent on creating a theocratic state, or the last defenders of religion and family values in America). Bringing the story of the religious right up to the Obama administration, this revised fourth edition explores the history of the movement in twentieth and early twenty-first century American politics. The authors review the expansion of the Christian Right through George W. Bush’s second administration and evaluate how the religious right fared in the 2006 and 2008 elections. Although figureheads of the religious right remain in the news, their power in Washington may be declining, and the authors consider the fate of the religious right under the Obama administration. Examining how the religious right both does and does not fit into the proper role of religious groups in American politics, Onward Christian Soldiers? is an essential addition to the Dilemmas in American Politics series.
Governors are the most prominent political actor in state politics and a subject of continuous study by scholars. An analysis of the research produced over the past five decades reveals a pattern of topics investigated as well as trends in interest level of those topics. A brief review of some of the findings is presented as well as suggestions for additional research. Finally, a proposal to rethink how we study and evaluate governors is offered.
The stability of abortion opinions suggests that pre-adult factors influence these attitudes more than contemporaneous political events. Surprisingly, however, we know little about the origins of abortion opinions, no doubt because the majority of research focuses on cross-sectional analyses of patterns across cohorts. We use a developmental model that links familial and contextual factors during adolescence to abortion attitudes years later when respondents are between 21 and 38 years old. Findings show that religious adherence and maternal gender role values are significant predictors of adult abortion opinions, even after controlling for contemporaneous religious adherence and the respondents’ own views on gender roles. Adolescent religious adherence matters more than religious denomination for adult abortion attitudes. The results have important implications for future trends in abortion attitudes in light of declining religiosity among Americans.
Full-text available
This paper explores the determinants of gubernatorial election outcomes at the national level. Gubernatorial elections are found to be sensitive to swings in the national economy and presidential popularity. Somewhat surprisingly, gubernatorial elections are not found to be sensitive to the type of midterm election phenomenon found in congressional elections. Also, incumbency is a very important determinant of gubernatorial election outcomes.
This book examines the shape and direction of public attitudes toward abortion. It looks at the social and demographic basis of public opinion on the abortion issue. The book is also concerned with the consequences of abortion politics.
Inference for logistic regression based on the information matrix may be poor. This is noted in two examples in which confidence regions are examined. A measure to detect such inadequacies is presented; it judges the quadratic approximation to the likelihood surface, which justifies the usual procedure.
As the U.S. states develop their political institutions and take greater responsibility for their economic well-being, two concerns that have long driven research on national elections—electoral insulation and economic accountability—should become central in research on state elections. I investigate institutionalization's effects on the vulnerability of state elections to major periodic forces—coattails, turnout, and economic conditions—and how political responsibility for economic growth is apportioned between presidents and governors in state elections. The investigation relies upon dynamic models of state legislative and gubernatorial outcomes estimated with a pooled data set comprised of most states and elections in the years 1940–82. The results, which have important implications for state government more broadly, indicate that institutionalization has substantially insulated legislative elections against major threats and that state legislators and governors have less to fear from their state economies than is often thought, but also that state elections are becoming more susceptible to swings in the national economy.
For tests of a single parameter in the binomial logit model, Wald's test is shown to behave in an aberrant manner. In particular, the test statistic decreases to zero as the distance between the parameter estimate and null value increases, and the power of the test, based on its large-sample distribution, decreases to the significance level for alternatives sufficiently far from the null value.
Gubernatorial and presidential elections over the period 1947–1986 are examined, using a previously reported process for decomposing partisan electoral outcomes series into their longterm and short-term components. These measures are employed to examine the proposition that gubernatorial elections have become increasingly isolated from outside forces. It is found that presidential coattails appear to be declining in importance (but not only because a number of states have moved to off-year elections). Gubernatorial elections have converged around a national pattern of relatively close competition, unlike state-level presidential contests, which have shifted in favor of Republican candidates. The pattern of gubernatorial outcomes varies more from state to state, however. In specific elections, the short-term forces remain in rough equilibrium between the parties in gubernatorial contests, but not in presidential contests, where the average short-term shifts favoring one party or the other fluctuate from one election to the next. On the other hand, gubernatorial elections respond less uniformly than presidential elections to these election-specific, national-level forces. This evidence suggests that the gubernatorial election contest has, in general, become more distinctive from the national context, reflecting a more fully autonomous office.
A number of recent attempts have been made to construct statistical models for predicting election outcomes. Such models have been applied to elections in various settings, though in the American context they have been limited mainly to elections for the U.S. House of Representatives, the Senate, and the presidency. The present study is an attempt to apply some variations of these models to gubernatorial elections in the United States in the period from 1910 to 1970.