University of Utah
The Impact of the Australian Ballot on Member Behavior in the U.S House of Representatives
Author(s): Jill N. Wittrock, Stephen C. Nemeth, Howard Sanborn, Brian DiSarro, Peverill
Squire
Source: Political Research Quarterly, Vol. 61, No. 3 (Sep., 2008), pp. 434-444
Published by: Sage Publications, Inc. on behalf of the University of Utah
Stable URL: http://www.jstor.org/stable/20299746 .
Accessed: 15/09/2011 11:09
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp
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 [email protected].
Sage Publications, Inc. and University of Utah are collaborating with JSTOR to digitize, preserve and extend
access to Political Research Quarterly.
http://www.jstor.org
Pol?tica] Research Quarterly
Volume 61 Number 3
September 2008 434-444
O 20?S University of Utah
The Impact of the Australian
in
Ballot on Member
Behavior
the U.S. House of Representatives
10.1177/1?65912907307291
http://prq.sagepub.com
hosted at
http://onlme.sagepub.com
Jill N. Wittrock
Stephen C. Nemeth
Howard Sanborn
Brian DiSarro
of Iowa,
University
Peverill
Iowa City
Squire
of Missouri,
University
Columbia
U.S. House of
and Sala linked the development of committee property rights in the late-nineteenth-century
as
to
of
the
Australian
ballot.
members
the
introduction
If,
they posited,
sought personal reputations
Representatives
to carry them to reelection in the new electoral environment, the current article argues that behaviors with more imme
diate political payoffs also should have changed in ways their theory would predict. The article examines whether
committee assignments, floor voting behavior, and the distribution of pork barrel projects changed in predicted ways
and finds supportive outcomes, but usually only when the office bloc ballot, and not the party bloc ballot, was in use.
Katz
U.S. House
Keywords:
of the more
institutional
One
of Representatives;
inventive
evolution
Australian
ballot; committee assignments; party voting; pork barrel projects
theories about Congress's
and Sala's (1996)
is Katz
of
the development
of committee
property
linkage
House
U.S.
of
in
the
late-nineteenth-century
rights
to
of
the
Australian
the
introduction
Representatives
ballot. Katz
and Sala argued that the Australian ballot?
ballot cast in secret that replaced
cast
in public?greatly
ballot
a government-printed
a party-produced
increased
the incentive
for members
of the House
votes. This,
pursue personal constituency
the rise of committee
property
rights
to
in turn, led to
as members
from term
assignments
sought to keep their committee
to term because of the potential electoral benefits they
derived from them.
Katz
While
seems
it
(1996) theory is believable,
to expect that an assertion of com
and Sala's
reasonable
rights was
not apt to be among House
property
to the new electoral process.
members'
first responses
After all, committee
property rights generally pay off
mittee
over the long run, particularly when associated
some variant of a seniority system (and, of course,
an intention to serve over a long period of time). If, as
were
of Congress
Katz and Sala suggest, members
only
with
motivated
to carry
to seek personal electoral benefits
in the new Australian
ballot electoral
them to reelection
434
then behaviors
environment,
payoffs
political
their theory would
also
with
more
have
changed
should
immediate
in ways
predict.
In this article, we examine whether
three different
sorts of everyday member behavior changed in hypoth
in the wake
of the adoption
of the
esized ways
we
test whether
the
Australian
ballot. Specifically,
floor voting
reform influenced committee
assignments,
behavior,
and the distribution
conjecture
Australian
of pork barrel projects. We
in states with
members
the
that House
pursued more
felt less compelled
ballot
prestigious
to vote with
committee
their party,
assignments,
than
and chased pork barrel projects more successfully
from states without
the reform.
their colleagues
We
take an additional
also
by Katz
and Sala (1996)
the Australian
of
impact
an important distinction
stated, states
either choose
office
bloc
adopting
a party
that taken
step beyond
the behavioral
by examining
two different forms,
ballot's
they did not explore. Simply
ballot
the Australian
column
ballot
ballot
had
to
or an
design
the office bloc
design. Arguably,
design was more
likely than the party column design
to change the electoral
incentive structure in the man
ner that Katz
lot reform
and Sala ascribed
generally
(Squire
to the Australian
et al. 2005).
bal
Wittrock et al. / The Impact of theAustralian Ballot
The Katz
Once
and Sala Theory
sarily
Katz and Sala (1996, 23) presented their theory
This
in a nutshell:
is our
argument
the
raised
changes
The
of
column
that would
arrangements
Congress
help them build personal reputations. Stable com
the leeway and
mittee assignments
give members
confidence
within
they need
to become
policy
experts
and Sala
a personal
saw committee
rights
and enhanced
of short-term
more
quickly
as con
a
for example,
observed,
the chief
object of the "party column"
is to facilitate
the voting of a "straight
ballot
that of the "Massachusetts"
ballot
ticket," while
the voters stop and think about each
is to make
Since
through
(1909, 259),
Ludington
the building
of personal
reputations,
as
a
in
issue area.
experts
given
specifically
policy
in this regard is a long-term
But reputation building
can
process. As we argue below, political
reputations
be established
bloc
Australian ballot's different variations (Allen 1910).
to
tributing
to an office
thus preferred by party organizations.
Politicians
and knowledgeable
observers at the time
understood
the potential electoral consequences
of the
reputation.
property
in 1897, and returning
ballot
was
a
of Congress. Hence,
domly selected members
to their same
incumbents
"norm" of reappointing
committees
would
be consistent with a wide
for building
to
four states had changed
while
the party column ballot became much more popular
than the office bloc ballot across the states, probably
because it resembled the old party produced ballots and
their committee jurisdictions. Policy
spread desire
design,
ballot in 1899 (Ludington 1909,260; 1911).Ultimately,
experts are better equipped to claim credit and are
more
takers on policies
noteworthy
position
than are ran
within their committee's
jurisdiction
variety
employed the office bloc ballot had switched to the
the office bloc ballot from the party column ballot.
Nebraska flipped back and forth, first adopting the
office bloc ballot in 1891, then switching to the party
ballot
of members
interest
in institutional
Katz
chosen, a state's ballot design was not neces
fixed.
Indeed, by 1905, six states that first
party column
succinctly.
435
in turn, it is natural that most of the States
which
have the former type of ballot provide
for the "straight ticket" voting and
specifically
office
that most
of those which
have
the latter do not.
behaviors.
Analysis of voting data from 1890 to 1908 confirms
thatballot type differences influenced theway people
behaved (Allen 1906; Rusk 1970, 1235).
and the Different
Personal Reputations
Ballot
Forms of the Australian
The difference
in ballots
is crucial
to the Katz
and
Sala (1996) theory because the office bloc ballot
ballot is treated as being uniform
it by Katz and Sala
that adopted
in the
There
variations
were,
however,
(1996).
ballot that have significant
design of the Australian
a state took
for
their
implications
theory. When
The Australian
across
the states
an official ballot,
it had
for producing
responsibility
to decide
labels
how names,
and
contests,
party
on it. Most,
would
be organized
but not all, of the
first
states
for
the
that adopted
the Australian
ballot opted
as the
also
known
bloc
design,
office
or blanket
Massachusetts
lot is designed
ballot.
The
so that the names
office
bloc bal
of candidates
are
more
ing legislators
tial incentive
reference
to this difference,
noting,
to evaluate and vote for candidates
allows
their respective
for various
offices
party name.
in a col
other. Katz
a passing
"A system that
each
and Sala made
voters
on
an office-by-office,
basis
encourages
case-by-case
to invest more
in their personal
incumbents
reputa
individual
of candidates
thus giv
substan
and from
tions
the names
for split ticket voting,
states using such ballots
often made
it
contrast, party column ballots, which
more difficult
to split ticket vote, gave legislators
less
reason to differentiate
themselves
from their parties
are contesting.
states adopting
the reform in
Most
later years
their ballots
party
using
organized
came
to
ballot
which
be
called
the
Indiana
columns,
umn under
from
as a way
to pursue personal
reputations
voters.
of distinguishing
themselves
before
the
In
grouped together under the title of the office they
(Ludington 1909, 260). Party column ballots place
allowed
easily
While
than when
voters
candidates
cannot
on
discriminate
a partisan
slate"
between
(p. 24).
they did not incorporate this distinction into
to
it seems clear that the incentive
analysis,
a
more
to
be
would
develop
personal reputation
likely
in states with office column ballots
than in
develop
their
states with
party column
ballots.
436
Political Research Quarterly
period allows us to investigate the effects of ballot type
on member
as the Australian
behavior
ballot gained
The Australian
Ballot, Personal
and Immediate
Behavioral
Reputations,
We think that in their efforts to distinguish them
selves
from other members
of Congress,
states with Australian
ballots
tives from
bloc
representa
of the office
behave
type would
leagues from other states
members
than their col
differently
in at least three ways. First,
to establish personal
reputations
motivated
to be more
on
in securing positions
aggressive
Second,
preferred committees.
they should feel freer
to deviate
from the party line in floor voting. And
ought
third, they should seek to bring more pork barrel pro
jects and dollars back to their constituents.
Assignments
the early history of the House,
speakers used
as
a
for
reward
assignments
loyalty to them
selves and to the party (Gamm and Shepsle
1989;
Jenkins 1998). As the number of standing committees
to
party leaders had more
opportunities
over
a
number
their
control
of
represen
solidify
larger
it was not until the adoption of the
tatives. Arguably,
increased,
ballot
that House
an incen
members
gained
to establish
tive to seek positions on certain committees
a personal reputation with their constituents.
from
states
of
committee
assignment
had. Party column
to
control
elected
that issued
from
beholden
a party column
to pursue
that office bloc
the same motivation
did not have
sentatives
spanned
minimizes
limit our
existence
under
study. This restriction
that might arise from the
any idiosyncrasies
of
short-lived
bodies or the dissolution
of com
pursuit
mittees. We draw heavily from Canon, Nelson,
and
(2006) historical work on standing commit
information available from
tees, as well as biographical
the Inter-university Consortium
for Political and Social
Stewart's
Research (ICPSR).
There
atically
a number
have been
classify
committee
of articles
hierarchies
that system
in the House.
Bullock and Sprague (1969) determined the rate of
into a committee
transfers
as a ratio of all transfers
in
ballots
repre
left a great deal
Congress.
were not as important for members
running on party
run
column ballots as they were for their colleagues
ning on office bloc ballots. With office bloc ballots,
ers directly chose the candidates
for each office
vot
and
could more
easily determine whom to reward or punish
in the voting booth. Representatives
elected with these
inclined to pursue committee
ballots should be more
remaining in office.
of
Our analysis
the Forty-ninth
them the resources
committee
through
Fifty-sixth
to gain
their chances of
assignments
Congresses.
committee
is.
assignment
a net
(1988)
In contrast,
transfer domi
Munger
suggested
using
nance measure,
are compared
where
committees
in
to
this approach
is
pairs. The logic behind
specifi
are
which
to
determine
committees
cally
preferred
a large-scale matrix and run
one would
of comparisons,
others. After
ning
have
developing
several
iterations
a long chain of hierarchies
with
the most pre
at the top and the least preferred
committee
at the bottom.
committee
ferred
Groseclose
and
Stewart
each committee
to the party leadership
for their place
in
name
Personal
and
reputation
recognition
give
favor and increase
the
ballot
members
party
organizations;
that used these ballots were
districts
assignments
their constituents'
able
a pre
the
that would
In their view, the higher
into a committee,
the more desir
Stewart
and
(1998;
a
more
to
intu
1999) developed
technique
itively assess the value of serving on a given commit
tee. Their "Grosewart index" assigns a cardinal value to
As noted above, the type of Australian
ballot used
for electing
is
in
legislators
understanding
important
of House members.
Those who were
the motivations
ferred
the entire period
the rate of transfers
During
committee
elected
the country. We
throughout
to only those committees whose
and out of that committee.
Committee
Australian
use
widespread
examination
Changes
covers
This
Groseclose
based on transfers. This
maximizing
valuation
of
a likelihood
a committee
member
each
function
is generated
by
the average
the members
who
to obtain
committee
by
served during the period under investigation.1
The fundamental
of this method
assumption
is that
on his
has
"property rights"
If he transfers onto a committee,
he stays
assignment.
on for as long as the committee
holds enough value
to assist him in his work.
If he transfers off, he does
so willingly.
A
transfer from a committee
there is a better
assignment
awaiting
leave a plum assignment
an even better one.
would
not
implies
the member.
unless
that
He
promised
Unlike
the previous measures,
the Grosewart
index
has intrinsic meaning. A positive value implies that the
to the member.
committee
is beneficial
In addition, we
can develop a quantitative understanding
of each com
mittee's relationship to one another. Consider amember
serving
on the Appropriations
Committee
(Grosewart
Wittrock et al. / The Impact of theAustralian Ballot
index = 2.61). This member would need to transfer to
at least two other committees,
(1.26) and
say Rules
to justify
and Harbors
the
Rivers
(1.43),
leaving
Committee.
Appropriations
score of zero connotes
A
indifference
toward
serv
a member
receives neither
ing on a given committee;
A negative
benefit nor penalty from the assignment.
score signifies
that the committee
is a burden to the
member.
tee
He would
than continue
This
to any commit
rather not belong
one.
service on the burdensome
could be because
of the time and effort
and a dearth
of service
of subsequent
benefits
required
to the
career.
member's
is the over
Our dependent
variable in this analysis
a
of
House
member's
committee
all value
portfolio.
the average values
(v.) of each committee
to cre
to in a given congress
that a member
belongs
ate the portfolio
value. Members
with high portfolio
sum
We
portfolio will be higher than those elected from
are, of course, other
as controls.
Canon
There
entered
variables
associated
the South
First,
of whether
and Stewart
Fifty-fourth
Congress,
Committee
Expenditures
For example,
2002).
he
let us
served
(Grosewart
the State
index = -0.04),
the Foreign Affairs Committee (1.78), and the Library
(1.09). His portfolio value would be 2.83,
In
well above the average portfolio
value of 0.974.
we
can
use
to
the
index
Grosewart
short,
compare
to one another and contrast the value of
committees
Committee
those
committee
given
congress.
assignments
across
members
in a
sentatives
mittee
used
on Katz
claim
that the
(1996)
a
the
of
Australian
ballot
had
adoption
positive effect
on the career-driven motivations
of House members:
focuses
and
Sala's
to account
tion on committee
who
seek
better
representative
the Australian
is elected
from
ballot,
portfolio will be higher
are elected from pre-Australian
mittee
Our
ment:
that uses
We
more
also
hypothesis
the type of Australian
cally the office bloc ballot, affect member behavior? Are
members
from states that adopted these ballots more
career-oriented
than those from
states that did not?
Hypothesis 2: During the period 1885 to 1901, if a
is elected from
representative
office bloc ballot,
a state that uses
then the value of his committee
an
There
is evidence
assignments.
were
had attended college
inclined
or
were
committee
assignments
committees
could provide
advantages.
look at a number
of variables
that control
of a given Congress,
could have influenced
a
As
assignments.
seniority norm became
with
institutionalized,
many members
longer
tenures
First-term
have
received
might
status is also included
new member
to the House
preferred
positions.
as a control variable.
might
not have
been
any particularly
plum assignments.
we
include measures
for the dis
Additionally,
tances from the chamber and party medians
based on
given
scores
DW-NOMINATE
and Rosenthal
(Poole
the
House
have
1997).
might
on preferred commit
been likely to gain membership
tees. The same might hold true for extreme members
of
to test our argu
ballot used, specifi
a college
of educa
beginning
committee
we
states.
include
Second, we
for the ancillary effects
on member
for organizational
constraints
behavior.
as
We conjecture
that seniority, measured
the cumu
lative number of years and months
in office at the
than those who
ballot
sought and been granted com
for reasons other than the ballot
them the most
of his com
is a means
second
does
a state
then the value
is elected
specific
able to assess which
Extreme
Hypothesis 1: During the period 1885 to 1901, if a
measure
from a former Con
have
might
assignments
in their states.
variable
A
test two hypotheses. The first hypothesis
We
characteristics.
is a dichotomous
state, entered to capture any residue of the
that region
much
of political
argument
explains
in late nineteenth
behavior
repre
century. Southern
(R-NY). During the
on
be
Stewart
federate
to
(Canon
and
states.
that must
variables
member
with
variable
a member
that members
consider Rep. Lemuel Quigg
ballot
(2002)
com
accounted
for a number of factors in assessing
we
mittee
hierarchies.
control for two
Accordingly,
values
House
and party column
pre-Australian
Those
to the most
committees.
belong
sought-after
to
members
with negative
belong
portfolios
are
to
in
the
committees
that
burdens
their service
437
members
of
the Republican
and Democratic
parties. Finally,
a representative
was
include a control for whether
a member
of amajority
that
controlled
less
than
party
55 percent
ties might
party's
We
of the House.
have
been
Leaders
with
tight majori
their
likely to secure
on desirable
committees.
more
representation
test our hypotheses
using a series of ordinary
least squares (OLS) regressions. Model
1 in Table 1
reveals that the Australian
ballot is statistically
signifi
cant in explaining committee
service.2 Our first hypoth
esis is supported; members
elected from a district using
the Australian
ballot possess
slightly more valuable
committee
portfolios
than
their
counterparts.3
Our
438
Political Research Quarterly
Table 1
Effect of Ballot Structure on Preferred
Committee
Least
Ordinary
Assignment,
had more
Squares
Model 2:
Office and
Party Ballots
1:
Australian
Ballot
South
?
320***
317***
.081***
.081***
median
chamber
(.130)
.020
.020
(.039)
(.039)
also
Office ballot
27^***
Party ballot
547***
005
039)
structure
554***
One
in parentheses.
committee
portfolio
.05. ***/? <
.001
bloc
We
find
is that it is the type of Australian
test this, we
the
separate
distinct
ballot
into
variable
types:
To
elected
sessed, on average,
those elected with
greater
model
a member's
and party column.
support inmodel 2 for our second
esis. Members
for
coefficient
of
standard
test).
(one-tailed
ballot
office
with
reported
value
variable:
Dependent
in each Congress.
however,
contention,
that matters.
ballot
Australian
are
coefficients
Unstandardized
with
0.18
portfolios
another ballot.
the office
than the coefficient
1. Thus,
office
not
only
bloc
hypoth
ballots pos
units greater than
bloc
The
size
is three
ballot
the
times
ballot
for the Australian
bloc
is the office
of
in
ballot
it has
but substantively
statistically,
highly significant
in pursu
a larger effect on the behavior of members
assignments.
ing committee
We
used
can conclude
office
bloc
that representatives
those
ballots?not
ballot
group of Australian
tive committee assignments.
from states that
from
the larger
attrac
more
states?gained
These representatives
now
of nineteenth-century
the House,
within
that favored
of
party
the core
America
government.
errors
*p <
century
246
.241
by representa
one of
represents
late nineteenth
culture
the political
institutions
political
050)
(.050)
Note:
and pork
century
partisan times in the nation's history and has
been referred to as the "golden age of party politics."
sense of party unity was produced by
The heightened
044)
Adjusted R2
N= 2,585
services
the most
(.034)
Constant
to provide
the level of party voting
influence
tives. The
.063*
ballot
Australian
enhanced
The introduction of the Australian ballot should
-.130
(.130)
from
They
Party Unity
(.040)
-.124
party median
Distance
of ways.
institutional
membership
to their constituents.
.131***
130***
from
mittees,
representatives
in a number
of reelection
(.033)
(.040)
Distance
com
joining more valuable
could increase their chances
j2i***
.123***
(.033)
Small party majority
By
committee
(.005)
(.005)
College
survival.
of electoral
as a means
district
influ
gaining
potentially
prestige,
ence over the internal functioning
of the House. More
could use their
representatives
important, however,
(.040)
(.040)
Seniority
from office
representatives
appeal to the voters in their home
their
(.049)
(.049)
Freshman
this new
with
from party
colleagues
level of accountability,
to
needed
bloc districts
their
than
states. With
a connection
to build
-.007
.022
?
constituents
their
column
(OLS) Regressions
Model
of an incentive
political
was faith
loyalty.
beliefs
America,
and a ballot
of nineteenth
in the benefits
of party
Parties,
during presidential
especially
were
to the
to bring cohesion
believed
elections,
and legislative
government
by linking the executive
were
branches
1988).
Representatives
(Sundquist
on their party platforms
and were
based
elected
it was
of their party while
judged by the performance
were
in power. Representatives
accord with their party's wishes,
their own personal
flicted with
to vote in
expected
even when they con
beliefs
(Cooper and
who rejected their party lead
1981). Members
were
seen
as
the mandate
given to
rejecting
ership
them by the voters (Jones 1968).
member
of Representatives,
the House
Within
Brady
was
speakers, color
by powerful
to as "czars"
1961). The
(Galloway
fully referred
over
the
committees,
presided
speaker
appointed
and had the ability to use a variety
Rules Committee,
to ensure compli
rewards and sanctions
of political
behavior
controlled
ance with
his prerogatives
1981).
(Cooper and Brady
who
In this atmosphere,
any representatives
sought to
own
or
their
their legislation
advance
legislative
careers had to remain in the speaker's
favor. Given
that one's
made
success was
it difficult
of independence
this
tied to the speaker's whim,
to assert any measure
for members
(Cooper
and Brady
1981).
Wittrock et al. / The Impact of theAustralian Ballot
Table 2
Party unity also was reinforced by the control exerted
and election
by political parties over the nomination
in large
electoral fortunes were,
processes. Members'
Party Unity Scores, Forty-Ninth
to Fifty-Sixth Congresses
part, under the control of their parties and various levels
of party officials.
In many
situations factional candi
to a
dates arose, making
the official candidate beholden
of
for
their
these
officials
range
support against
party
factions. The candidate also needed to appeal to party
439
Average Unity (by Party)
Congress
Average Party Unity
Democrats
Republicans
49
82.38
50
83.71
51
91.39
52
78.33
53
88.03
54
84.52
55
92.31
56
92.37
78.94
77.70
88.20
74.84
86.79
87.88
90.63
89.91
86.84
90.27
94.31
88.04
90.17
83.11
93.36
94.49
for their help on election day, such as paying for
individual ticket peddlers to ensure the correct ballots
officials
were distributed (Reynolds andMcCormick 1986).The
of voting during the era was characterized by an
and spectacle rather than
atmosphere of auctioneering
process
civic responsibility (Rusk 1970). Peddlers, mavericks,
and others provided a variety of ballots, party-approved
and otherwise,
and opportunities
for voters to deviate
from
their party's wishes.
Those
that received
their
ballots
a distinctive
form, on paper
color or size than those of other par
received
party's
either of different
ties. In the best case scenario, a party's ballot was openly
each individual's
cast, making
political
preferences
to party workers
obvious
and other voters. While
the
election process itself did not cement the loyalty of the
candidate to the party leadership, he still needed their
support to overcome various threats that could arise dur
ing the course of a campaign.
The adoption
structure
of the Australian
ballot
should
have
allowed
members
of the House
to exer
cise greater
if there was
and
from the party, particularly
independence
a conflict between
the views of the party
even if
the member's
Moreover,
constituency.
were con
preferences
needed to develop per
to remain in office under office
This
score
is a percentage
the number of
representing
of the House votes with his party on
times a member
votes
by the total number of
on per
votes
the member
voted
party
support
are
as
votes
defined
those in
Party support
Congress.4
which at least 50 percent of one party vote against at
party
support
divided
least 50 percent of the opposing party. For our partic
were noted for their high
time period, members
ular
support
analysis have
in the
scores; very few individuals
scores lower than 50 percent. Also of
note
higher
party
is
the
Republican
party,
who
score
noted
(1979),
Hurley
for members
in line with
of
Brady, Cooper,
the more centralized
the
and
sys
tem of leadership typical of that party during this
time period.
Table
2 shows
the degree of party unity
this time period.
and by party over
and constituency
policy
gruent, members
increasingly
by Congress
sonal
a state has adopted the Australian
ballot structure. For Hypothesis
two
4, we examine
one
variables
that indicate whether
the state adopted
party
constituencies
ballot
(1987)
which
All
In Cain,
and Fiorina's
systems.
Ferejohn,
a "personal vote,"
term, they had to establish
actions.
typically requires independent
of this suggests
the following
hypotheses:
representatives
Australian
ballot
scores
ballot
will
states
that
have
lower
than representatives
states.
the
adopted
party unity
from pre-Australian
from
states
that
the
adopted
office bloc ballot will have lower party unity
scores
ballot
We
party unity
using
(Brady, Cooper,
of the Australian
ballot:
the office
or the party column ballot. We
include
control variables
in the equations. We add
appropriate
a variable that indicates how
long the representative has
in the House,
that members
with
conjecturing
more
on
to
service
be
may
greater
likely
rely
voting
cues other than party. Second, we also hypothesize
the
served
representatives may be more likely
cues
cues
rather than other possible
party
we
and
Judd
include
(Weinbaum
1970). Consequently,
a variable indicating first-term status. We enter a vari
the standard measure
and Hurley
1979,
to follow
able for the margin of victory in the general election for
each representative,
testing whether more comfortable
victories
afford members
from
greater independence
than representatives
from pre-Australian
and party ballot states.
test our hypotheses
ballot
inverse?first-term
Hypothesis 4: During the period 1885 to 1901,
representatives
3, the independent variable of
is whether
of the two variants
bloc
Hypothesis 3: During the period 1885 to 1901,
from
For Hypothesis
interest
of
383).
party
period
leaders.
employed
Because
amore
Republicans
centralized
during
leadership
this time
system
440
Political Research Quarterly
Table 3
Effect of Ballot Structure on Party Unity
Scores, Grouped Logit Analysis
(Cooper and Brady 1981) that resulted in higher party
in Table 2), we include a
unity scores (as indicated
dichotomous
include
variable
also
party. We
the party median
from the chamber median based on DW
controls
and distance
for political
from
for distance
NOMINATE scores (Poole and Rosenthal 1997). A
variable indicating small majorities (those under 55
is included
because
we
suspect that smaller
more supportive of
to
members
be
prompt
majorities
their party. A control for each Congress
is entered,
over
because
time members
of Congress
adapt to the
percent)
political
landscape of new ballot rules and
their
behavior
within
the House
accord
change
changing
may
ingly. Finally, we again
to control for the South.
add a dichotomous
variable
the variable is
particular statistical method used. Because
constructed as a fraction, OLS will provide biased coef
ficients and standard errors (Smith 2001). Another con
cern is that our dependent variable ismade from groups
that vary in size for each
(Democrats and Republicans)
means
This
that OLS assumptions
particular Congress.
about constant variance in the error terms are violated
(Binder 1999).As a result, grouped logitwith weighted
least squares estimates to account for heteroskedasticity
to employ.
is the more appropriate model
As indicated by the results in Table 3, we find mixed
The coefficient
for the
support for our two hypotheses.
reform, while
in the predicted
direc
is not statistically
significant.
Representatives
states with office bloc ballots, however, demon
strate lower party unity scores than representatives
tion,
from
have a lower average score?76.8
the baseline of 79.8 percent.
percent
compared
to
provide a variety of substan
tive results. Following
the results of Table 2, we find
that party has a positive
and statistically
significant
were more
effect on party unity as Republicans
likely
control
variables
to have higher party unity scores than their Democratic
service resulted in more,
counterparts. Longer House
not
less, party
unity
and first-term
accounts
not
status, while
not
for less,
more,
significant,
statistically
party unity. These results are the reverse of our orig
The South exhibits a strong and pos
inal contention.
do
itive effect on party unity scores. Small majorities
a majority
has less
the expected
effect; when
than 55 percent of the total seats, we see that party
have
South
.674***
.685***
(.041)
Freshman
(.041)
-.016
-.013
(.035)
(.035)
.014***
Seniority
014***
(.003)
(.003)
-.0009
Margin
-.0008
(.001)
Distance
(.0007)
195***
from
party median
Distance
from
chamber
295***
(.037)
.034
.029
CUD
( 111)
-.030
median
-.030
(.030)
(.030)
.700***
Party
.706***
(.032)
(.032)
.070***
Congress
.072***
(.010)
Australian
ballot
(.010)
-.063
(.042)
?
Office ballot
178***
(.052)
Party ballot
.006
(.046)
Constant
1.00***
Adjusted R2
N= 2,496
Note:
from party column and pre-Australian
states, consistent
with Hypothesis
4. Predicted
generated
probabilities
from model
lend further support.5
2, while
small,
from states with these ballot reforms
Representatives
The
Model 2:
Office and
Party Ballots
(.037)
is our depen
party unity for each Congress
must
dent variable, special consideration
be given to the
ballot
1:
Ballot
Small party majority
Because
Australian
Model
Australian
Unstandardized
reported
***/? <
unity
with
standard
.001
(one-tailed
increases.
party median
.986***
(.043)
(.043)
.256
.260
logit
estimates
are
for distance
from
the
least squares
weighted
errors in parentheses.
test).
Coefficients
and chamber
median
are
statistically
insignificant.
There are two potential
for the rela
explanations
tively subdued impact of ballot reforms. One is that
of the House
parties were so central to the functioning
that their power over member behavior largely survived
the imposition
of new ballot rules. One particularly
comment
to this effect was made by Rep.
suggestive
Jacob Fassett, a New York Republican,
in 1910, a full
our
after
the
of
in
end
the period
decade
analysis:
are robust partisans, every one of us_I
it that no Democrat was elected to cooperate
We
take
with
our party nor was any Republican
elected to hand
over the Republican
to our
control of this House
... A man
to
have
opponents.
political
ought
Wittrock et al. / The Impact of theAustralian Ballot
my judgment, the
opinions and convictions_In
to
of opinion on unim
place
adjust differences
and on important questions of
portant questions,
...
and
party policy is not in public
public policy
but in the family caucus. (Jones 1968, 630)
ballot reform
the impact of the Australian
to
the
institutional
import given
by
swamped
in
that occurred
the House. The changes
party within
of the "revolt" against Speaker Cannon,
the wake
such as the removal of the speaker from the Rules
the election
to the Rules
of members
Committee,
to discharge bills
the ability of members
Committee,
and the election of all standing com
from committee,
of
mittees,
probably meant more for the development
than did the introduction
representatives
independent
of the Australian ballot (Cooper and Brady 1981,
So
415-16).
important
in essence,
while
to Progressives,
vote" oriented
"personal
have been more
ballot
reform
the creation
representative
the result of internal House
than of a shift in external
was
the more
of
well
may
changes
electoral
procedures.
around
revolves
second potential
explanation
the creation of new ballot rules not as an
The
viewing
antiparty reform,
the full blessing
but rather a reform undertaken
and
of
support
the major
with
parties
(Reynolds andMcCormick 1986;Ware 2000;Walker
In effect,
2005).
this is a nuanced
argument?corrupt
the adoption of Australian
party organizations
spawned
and
ballot rules (the classic
thesis) in the Northeast
elections
led to
Midwest
and volatility in congressional
their
in
the West.
adoption
states to the union
Republican
The
desire
and concern
to admit
over fusion
tickets inNorth Dakota and South Dakota led to the
adoption of the Australian ballot by Republicans in
those areas (Walker 2005). This behavior, however, did
or to theWest.
not remain confined to Republicans
In
in several con
the South, the presence of Republicans
gressional
to adopt
resulted
seats led Democratic-controlled
the Australian
ballot. While
in the reduction
of the fortunes
tion party, it also produced a decrease
as well.
tion of other party factions
members
elected
this
under
statehouses
this not
only
of the opposi
in the mobiliza
system
Consequently,
more
were
and more
likely to exhibit the
votes. This might
the
for
certain
party required
loyalty
help to account for limited findings here.
amenable
to party control
claiming
for pork barrel projects
Wilson
reported these particular appropriations
six
times greater than those obtained for
usually
appro
agriculture projects and even exceeded military
(1986)
priations for forts and fortifications. From 1888 to
term, the average repre
1901, in a given congressional
sentative in the House
received 3.3 river and harbor
projects with each being worth about $154,000. The
passed by the Rivers and Harbors Committee
involved
the dredging
of waterways
and the
usually
reinforcing of river banks and thus were identified with
projects
could easily claim credit
specific districts. Members
these projects to their constituents, mak
for delivering
targets for representatives
seeking
ing them appetizing
to establish
personal
is another
behavior that might have been influenced by the
reputations.
The Katz and Sala (1996) theory would suggest
that members
from Australian
states
ballot
should
have been more motivated
districts
than
to bring back pork to their
their colleagues
from non-Australian
states. Given
ballot
the differences
bring federal money
enhance
motivated
to
to their district in order to
their reelection
test
To
types, we
in the office
in ballot
argue that members
seeking reelection
bloc ballot states ought to be the more
prospects.
we use
notions,
these
data
collected
by
Wilson (1986) on river and harbor projects from 1888
to 1901. The
information
projects
projects
set contains,
among other items,
of river and harbor
the
number
regarding
data
by a district and the dollar share of
received by a district. For the purposes of our
all river and harbor projects are included.
received
analysis,
Our hypotheses
harbor
about
the distribution
of river
and
are straightforward:
appropriations
Hypothesis 5: During the period 1888 to 1901, if a
representative is from a state that has theAustralian
then his district will
in
receive more
ballot,
for river and harbor projects than
appropriations
the district
of a representative
ballot.
from a state with
the pre-Australian
Hypothesis 6: During the period 1888 to 1901, if a
is from a state that has an office
representative
bloc ballot,
then his district will receive more
river and harbor appropriations
than would
the
district of a representative
from a state with a
party column
Pork Barrel Expenditures
Credit
for
ballot. Appropriations
adoption of the Australian
river and harbor projects were a major federal budget
outlay during the last quarter of the nineteenth century.
were
In effect,
was
441
Our
focus
or pre-Australian
on the number
following hypotheses:
ballot.
of projects
produces
the
442
Political Research Quarterly
Hypothesis 7: During the period 1888 to 1901, if a
Table 4
Effect of Ballot Structure on Pork Barrel
the
is from a state that has
representative
then his district will receive
Australian
ballot,
more
river
projects
from a state with
of a representative
Australian
ballot.
The most
Model 2:
Model 4:
Model 1 : Office and Model 3: Office and
Australian
Australian
Party
Party
Ballot
Ballots
Ballot
Ballots
a series of OLS
Party
Seniority
test of these hypotheses
is through
pork barrel data
regressions. Wilson's
direct
(1986) contain information at the district level regard
the dollar
ing the number of river and harbor projects,
share of the projects obtained by the district, and infor
from the district.
about the representatives
mation
the late nineteenth
century, the House River
During
and Harbor Committee
reported one omnibus bill and
bills every Con
large project
appropriation
on the
the
Most
of
gress.
maneuvering
political
omnibus and large project appropriation bills occurred
.183*
.193*
.084*
.088*
(.088)
.001
(.088)
(.039)
(.039)
(.008)
Committee
Margin
Australian
the River
were
We
and Harbor
These
Committee.
bills
rarely amended after leaving the committee.
two series of models
for both
constructed
the
ballot
and the number of projects.
of appropriations
and projects,
The dependent
variables,
appropriations
data for river and harbors pro
reflect appropriations
period and refer to the
jects during a thirteen-year
a congres
obtained
during
pork barrel quantities
and
sional session. The natural log of appropriations
the
of projects was used to normalize
the number
variables.
dichotomous
We
focus
Adoption
variable
of the Australian
was
5 and 7.
in Hypotheses
in Hypotheses
6 and 8 and
type
measures
for office
bloc
and party
variables were
control
Appropriate
added, among them party affiliation,
seniority,
and margin of victory.
mittee
assignment,
column
the
used
on ballot
use dichotomous
ballot
ballots.
com
are
estimates
for the four models
parameter
testing the rela
reported in Table 4. For our first model
Australian
ballot
of
between
the
the
presence
tionship
The
and the amount of appropriations, we find that adoption
of the Australian
ballot is statistically
significant and in
and in
also
is
the correct direction.
Party
significant
direction. And as Wilson
the expected
(1986) found,
(Rules, Appropriations,
being on a power committee
and in the
is statistically
and Means)
significant
Ways
predicted
direction.
variables?margin
The
of victory
explanatory
remaining
not
and seniority?did
-.002
-.002
(.003)
(.008)
(.003)
.208***
.591***
.208***
(.117)
(.117)
(.052)
(.052)
.000
.000
.000
.000
(.000)
(.000)
(.000)
(.000)
.285***
.080*
(.002)
(.042)
Office ballot
299**
.124**
(.116)
Party ballot
Constant
(.052)
.211*
.044
(.101)
(.045)
.706***
10.357***
10.382***
(.094)
(.093)
(.042)
(.042)
.026
.024
.021
.022
Adjusted R2
N= 1,355
coefficients
are
in parentheses.
of dollars
formation
*p <
Dependent
or projects.
.01. ***p < .001
variable
(one-tailed
Note:
Unstandardized
errors
amount
.001
.590***
several
within
Projects
Expenditures
is from a state that has an office
representative
then his district will receive more
bloc ballot,
than would
the district
river and harbor projects
a party col
Least
Ordinary
Squares (OLS) Regression Analysis
the pre
Hypothesis 8: During the period 1888 to 1901, if a
from a state with
of a representative
umn or pre-Australian
ballot.
and Projects,
Expenditures
than the district
and harbor
.05. **p <
709***
with
reported
is the natural
standard
log
trans
test).
a statistically
impact on how much
significant
a representative
for his district. The
received
have
money
for
ballot on appropriations
impact of the Australian
with
is consistent
the River and Harbor committees
Katz
and Sala's
Once
again,
and
assignment
(1996) theory.
as indicated
in model
matter
2, committee
in determining
how
in appropriations.
More
party
receive
representatives
we
find that both the office bloc
however,
important,
are significant
and
and party column ballot variables
in the expected
direction. The office bloc coefficient
much
is, however,
office
bloc
larger. Therefore,
representatives
states received more
river and
from
than representatives
appropriations
umn and pre-Australian
ballot states.
set of models
from
harbor
party
col
the impact of
as
the specific
reform,
reforms of office bloc and party columns ballots, on
that the
3 indicates
the number of projects. Model
The
second
the Australian
ballot
between number
relationship
ence of the Australian
ballot
estimates
as well
of projects and the pres
is significant
and in the
Wittrock et al. / The Impact of theAustralian Ballot
direction.
The
hypothesized
party and a seat on a power
results
also
show
Notes
that
are statisti
committee
1. The
cally significant and in the hypothesized direction.
While
443
likelihood
as follows:
is constructed
function
not a substantial
change from representatives
in pre-Australian
that
states, these results
suggest
members
under
Australian
ballot
rules
operating
secured more
leagues
for their district
projects
from pre-Australian
ballot
T
n*
than their col
f?i*;i
states.
The fourth model accounts for differences in the
number
of projects received by representatives
district. The results are similar to those inmodel
for their
where T is the total number of transfers in the data set, J is the
number of committees in existence during the period,
vj is the
3. Party
direction. As
is statistically significant and in a positive
in the previous results, committee
is highly significant
and is in the positive direction.
Importantly, however,
but not the party column
the office bloc coefficient,
is statistically significant.
coefficient,
What do these results suggest? The Katz
of committee
action
j, and xj is the member's
to
at
t
transfer
transfer
0: no
off,
(-1:
regard
committee./
1: transfer on), a = 1 is assumed;
the denominator
is the
transfer,
number
and Sala
the Ways
and Means,
Australian
ballot
tinuous
however,
reputations. Our general findings,
an important codicil
to the Katz and Sala
in states using the
theory: But it was often members
office bloc form of the ballot?not
from
members
personal
confirm
were more apt
states using party column ballots?who
to gain preferred
to deviate
committee
assignments,
from voting with their party, and to bring home federal
dollars for pork barrel projects.
These findings are important because they constitute
substantial
support for the basic reasoning underpin
ning the Katz and Sala (1996) theory. Evidence that
members
behavior
altered
immediate
in predictable
ways in response
environment
lends credence
ing electoral
that they also
behavior
their more
altered
by asserting
their more
committee
legislative
to the chang
distant
to the notion
legislative
property rights.
committees,
not
significant.
who
have
for members
votes.
support
or ordinal
and nominal
variables
in this case was
with
however,
a Northern
Democrat
to the party
close
relatively
and 0.046,
(0.005
the analysis
variable;
was
and who
respectively).
as a log-transformed
expenditures
the parameter
estimates
did
not
significantly.
References
adjusted
in ways
to build
medians
reran
6. We
efforts
Rules
were
models
calculated
ten party
of service
and chamber
Allen, Philip Loring. 1906. Ballot laws and their workings.
Political Science Quarterly 21:38-58.
consistent with the theory developed by Katz and Sala
with
are
member
baseline
3.4 years
change
consistent
and/or
the office-bloc
to their mean
variables
lot). The
with
Conclusion
in activities
all
to theirmode and then changing a variable of interest (office bal
dependent
to engage
because
5. The predicted probabilities were calculated by setting con
of Representatives.
(1996). Members
operating under the reformed elec
toral rules were more likely than their other colleagues
and
unity
in at least
participated
than their party column ballot colleagues. This suggests
that not only the presence of the ballot, but also the type
to pork barrel legislation
of ballot, matters
in the late
to our findings,
representatives
on routine legislative behavior
the analysis
throughout
Appropriations,
scores
4. Party
from office bloc ballot states
receives.6 Representatives
not much more money?
receive more projects?but
According
their behavior
tests
t.
in transfer
involved
one-tailed
is the focus of Katz and Sala's (1996) work. Both the
which
for pork barrel behavior on the
Rivers and Harbors Committee.
Ballot type also makes
a representative
a difference
in how many
projects
House
use
hypotheses indicate a specific sign to the slope of the coefficients.
3.We attempted to predict the likelihood of being assigned to
ballot matters
nineteenth-century
of committees
2. We
(1996) theory is correct in that the presence of the
Australian
valuation
average
with
-.
1910. The multifarious
Review
Sarah
Binder,
Australian
ballot.
North
American
191:602-11.
1947-96.
A.
1999.
American
The
Political
Brady, David W,
Joseph
The
decline
of party
dynamics
Science
of legislative
gridlock,
Review
93:519-33.
and Patricia A. Hurley.
1979.
Cooper,
in the US House
of Representatives,
1887-1968. Legislative Studies Quarterly 4:381-407.
Bullock,
Charles,
committee
and John
reassignments
1969. A research note on the
Sprague.
of Southern Democratic
Congressmen.
Journal of Politics 31:493-512.
Bruce
Cain,
E.,
John A.
and Morris
Ferejohn,
vote:
The personal
service
P. Fiorina.
1987.
electoral
inde
and
Constituency
MA: Harvard
pendence.
Cambridge,
University
Garrison
and Charles
Canon,
David,
Nelson,
Historical
Congresses,
set available
Canon,
David
congressional
1789-1947:
Stewart.
1st
and
theories
of
in
assignments
legislative
the annual meeting
Chicago.
79th
Data
hier
the U.S.
organization,
of
2006.
to
49-56.
at http://dspace.mit.edU/handle/1721.l/18171.
Stewart
III. 2002. Committee
T., and Charles
archies
Association,
committees,
standing
House
of Representatives
Press.
Congress:
Testing
at
1789-1946.
Presented
the Midwest
Political
Science
444
Political Research Quarterly
Cooper,
and
W.
and David
Joseph,
leadership
style:
American
Science
Political
Gamm,
islative
and Charles
Tim,
seats
committee
American
in
Stewart
the
III.
United
of Political
Journal
of
Emergence
in the House
leg
and
Science
The
1998.
States
value
of
Jones,
Legislative
Charles
O.
1968.
G.
Joseph
Cannon
of
Journal
and Brian
of Politics
R. Sala.
Howard
W.
and the electoral
assignments,
Science
Review
90:21-33.
30:617-46.
connection.
Political
United
States.
1911.
University
American
American
of New
Munger, Michael
York
C.
Poole,
Keith
economic
Science
Review
1880-1910.
laws,
queues
sity Press.
John E,
Reynolds,
State
of desirable committee
versus
committee
Science
32:317-44.
of Political
1997. Congress:
T., and Howard Rosenthal.
history
3:252-61.
Albany:
Press.
1988. Allocation
Extended
assignments:
American
Journal
Political
ballot
of roll call
and Richard
voting.
New
York:
L. McCormick.
seats
in
era
of
expansion.
A political
Univer
Oxford
1986. Outlawing
"treachery":Split tickets and ballot laws inNew York andNew
Jersey, 1880-1910. Journal of American History 72:835-58.
coalition
turnout.
American
Groseclose.
the United
States
in the United
government
Science
Quarterly
Lee D. 2005.
role of the Australian
ballot
Political
reform
Science
Journal
of Political
1999.
The
value
of
1947-91.
Senate,
Alan.
2000.
States.
Political
103:613-35.
The
as a party-system
switch: The
in party-system
and devel
change
ballot
ballot
opment in theUSA. Party Politics
Ware,
Ludington, Arthur. 1909. Present status of ballot laws in the
-.
on
III, and Tim
Charles,
Walker,
American
the Australian
American
45:700-706.
committee
committee
1996. Careerism,
of
American Journal of Political Science 43:963-73.
Sundquist, James L. 1988. Needed: A political theory for the new
Smith: An essay on the limits of leadership in the House of
Representatives.
Jonathon N.,
Katz,
effect
1876-1908.
Keith E. Hamm,
Ronald
D. Hedlund,
and Gary
Peverill,
Squire,
Moncrief.
2005. Electoral
and
reforms, membership
stability,
the existence
of committee
in
state
American
property
rights
Stewart,
and
The
legislatures. British Journal of Political Science 35:169-81.
42:453-74.
rights and the emergence
in the nineteenth-century
23:493-519.
Quarterly
Studies
Science
1947-91.
House,
1998. Property
Jeffrey A.
committee
dominance
standing
Jenkins,
House.
races
and candidate
1989.
Shepsle.
committees
Standing
Senate, 1810-1825. Legislative Studies Quarterly 14:39-66.
Groseclose,
1970.
Smith, Mark A. 2001. The contingent effects of ballot initiatives
of Representatives.
Crowell.
institutions:
G.
split ticket voting:
Review
64:1220-38.
75:411-25.
and Kenneth
Gerald,
Jerrold
Rusk,
on
to Rayburn.
Cannon
the House
of
context
Institutional
from
Review
1961. History
Galloway,
George.
New York: Thomas
1981.
Brady.
House
The
Anti-partism
and
11:217-41.
party
control
of
political
reform in theUnited States: The case of theAustralian ballot.
British Journal of Political Science 30:1-29.
Weinbaum,
Marvin
G.,
and Dennis
R.
Judd.
1970.
In search
of a
mandated Congress. Midwest Journal of Political Science
14:276-302.
Wilson, Rick K. 1986. An empirical test of preferences for the
political pork barrel: District level appropriations for river and
harbor legislation, 1889-1913. American Journal of Political
Science
30:729-54.