Us Senators Presidents Bcburden
Us Senators Presidents Bcburden
Us Senators Presidents Bcburden
Presidential Candidates
BARRY C. BURDEN
Robert Dole was apparently more comfortable and successful as
a senator than as a candidate for president of the United States. There was real
concern within the Dole campaign in 1996 that his bid for the presidency was
hindered by his position as a senator. The potential liability of his office became
so great that Dole chose to abandon his seat and position as majority leader of
the U.S. Senate. This was done to redefine himself as a citizen and presidential
candidate rather than as a member of the Senate and to create more time for
campaigning. It is not clear that this strategy worked, however, as the DoleJack Kemp ticket was soundly defeated with only 41 percent of the popular
vote and less than a third of the electoral vote. Though his senatorial experience
surely did not cause his defeat, this story raises the question of whether candidates generally find Senate service a liability in their quests for White House.
Contrary to Doles view, the conventional wisdom holds that senators are
prime presidential material. The Senate is thought to be the major launching
pad for presidential contenders.1 It has been called the Mother of Presidents, presidential incubator, the presidential nursery, and presidential
pre-school.2 Nearly every senator has been considered a potential candidate
for president at one time or another simply because of the office he or she occu-
Stephen Hess, The Ultimate Insiders (Washington, DC: Brookings Institution, 1986), 70.
Helen Dewar, Senates Club of Failed Ambitions, Washington Post, 10 March 1980; Robert
L. Peabody, Norman J. Ornstein, and David W. Rohde, The United States Senate as a Presidential
Incubator: Many are Called but Few Are Chosen, Political Science Quarterly 91 (Summer 1976): 237
258; Neil MacNeil, The Presidential Nursery, Harpers Magazine, January 1972, 4850.
2
81
pies.3 Some elections have even seen hordes of senators running for president
simultaneously. For example, the 1976 contest brought out Senators Henry
Scoop Jackson, Frank Church, Birch Bayh, Lloyd Bentsen, and Fred Harris,
while the 1988 election featured Senators Joseph Biden, Bob Dole, Al Gore,
Gary Hart, Ernest Fritz Hollings, and Paul Simon. This strong presence
would not surprise most scholars and journalists, since they assume that senators are natural candidates for the White House. Among other assets, they have
demonstrated campaign experience, are able to raise large sums of money, enjoy wide name recognition, and have already managed to represent large and
often diverse state constituencies. In short, senators ought to have a head start
over other kinds of candidates when the presidential race begins.
Despite these clear expectations, senators have not often been successful
presidential candidates, particularly in the modern nominating era. The failure
of senators to meet expectations cannot be explained by a series of unconnected anecdotes; statistical evidence shows that the poor performance of senators
is more than coincidence. Before turning to empirical analysis or suggesting
explanations, I begin by reanalyzing the historical record to reject the basic contention that senators are ideal presidential candidates. The second section reviews relevant political science theory and considers two specific historical
events that might have affected senators success rates. The next section develops the notion of candidate investment found in ambition theory and suggests
four explanations flowing from it that might explain why governors in particular
perform better than senators. Finally, I collect data on all major presidential
contenders over the last forty years to determine how career backgrounds in
the Senate affect campaign success.
Hess, The Ultimate Insiders; Jonathan S. Krasno, Challengers, Competition, and Reelection: Comparing Senate and House Elections (New Haven: Yale University Press, 1994).
TABLE 1
Last Offices Held by Presidential Contenders and Nominees (18681972)
Major Party
Nominees (%)
Contenders
(%)
Nomination Success
Rate (Nominees/Contenders)
Lubalin
POR
POR
Computed by author
Vice President
Other or None
Governor
Federal Appointee
Senator
Representative
President
N
5.5
13.9
33.2
16.7
13.8
5.5
11.1
36
5.4
24.4
37.9
13.5
13.5
5.4
37
2.0
15.7
28.4
13.7
30.4
9.8
102
1.00
.56
.48
.36
.16
.02
.36
Note: Peabody, Ornstein, and Rohde (POR) do not include current or former presidents in
their calculations.
however. First, there are probably differences between former and sitting senators. Indeed, many successful presidential candidatesfrom Martin Van Buren
to Harry Trumanare former senators who worked their way to the presidency through the vice-presidency. Thus, the safest path from the Senate to the
White House might be indirect. Roughly 142 major-party nominees have run
for president, but just forty-four of these were ever in the Senate, most of them
earlier in their political careers. Former senators not only make up the lions
share of this group but also won more often. There is either a liability associated
with those who are currently senators or a benefit associated with having left
the Senate in the past, perhaps attributable to passing through the vice-presidency first, a sort of back door strategy.
The second refinement to the conclusion is to point out that governors are
senators main competition. Thus, the story is as much about governors successful presidential campaigns as it is about senators unsuccessful ones. Over
the last thirty years, senators accounted for the largest group of declared presidential candidates but governors were not far behind.4 Nearly two-thirds of
contemporary presidential candidates are drawn either from the Senate or governors offices. Longer-range data support this contention too. Table 1 presents
data on presidential nominees and the larger pool of presidential contenders
assembled by Eve Lubalin and separately by Robert L. Peabody, Norman J.
Ornstein, and David W. Rohde (hereafter POR) for the period covering 1868
to 1972.5 Data from both studies are presented to show that results are similar
despite their use of slightly different coding procedures. The POR data are further manipulated to compute success rates.
4
Paul R. Abramson, John H. Aldrich, and David W. Rohde, Change and Continuity in the 1998
Elections (Washington, DC: Congressional Quarterly, 1998).
5
Eve Lubalin, Presidential Ambition and Senatorial Behavior: The Impact of Ambition on the
Behavior of Incumbent Politicians (Ph.D. dissertation, The Johns Hopkins University, 1981); Peabody, Ornstein, and Rohde, The United States Senate as a Presidential Incubator.
The first column in Table 1 lists the last office held by presidential candidates. This is nearly the same as Joseph Schlesingers definition of a manifest
office.6 While not an ideal way to characterize candidates office-holding histories, it is reasonable shorthand at this point for suggesting from where party
candidates are drawn. The Lubalin and POR data roughly agree that governors
provide the largest number of major party nominees. Senators and other federal appointees such as cabinet secretaries each comprise about 14 percent of
the nominees, while other types of candidatesprobably those with military
and business backgroundsare a slightly larger proportion of the pool. For
now, note that this low number of senators as nominees is surprising given that
senators far outnumber governors and visible executive branch appointees
combined.
The fourth column of the table considers all presidential contenders, a
larger set of candidates that includes the eventual nominees. Contenders outnumber nominees almost three to one using the POR definition. Among contenders the position of senators changes dramatically. Senators are one of the
smallest groups of nominees despite being the largest groups of contenders.
Governors, in contrast, tend to be contenders just as often as senators but are
much more likely to win nomination.
More interesting for our purposes is the rate at which contenders become
nominees. The final column of Table 1 presents the nomination success rate
(NSR), computed as the number of nominees divided by the number of contenders. The mean NSR for all contenders is .36, indicating that about one of
every three contenders wins nomination. If senators, governors, and other office-holders had equal success rates, all of the entries in the final column would
be nearly equal and close to this value. They clearly are not. Aside from House
members, who are few in number, the lowest value is for senators. Only three
out of every twenty senators who run for the presidency will be nominated.
In contrast, nearly half of governors who run for president win their parties
nominations. Governors nomination rate is nearly three times that of senators.
The simple evidence presented so far demonstrates that senators have fared
less well than conventional wisdom would suggest. Their failures have opened
the door for governors. Despite this clear statistical pattern, it is tempting to
explain some recent failures of senators using stories specific to particular elections. For example, one might argue that ex-Senator Dole lost the 1996 presidential election because he was perceived as too old for the job or because his
opponent benefited from presiding over a booming economy. Several other
senators lost in November because they were thought to be too ideologically
extreme (Barry Goldwater in 1964, George McGovern in 1972, and Walter
Mondale in 1984). One might argue that former Senator Hubert Humphrey
lost in 1968 because of a fluke third-party candidacy, a lack of unity within the
6
John H. Aldrich, Before the Convention: Strategies and Choices in Presidential Nomination Campaigns (Chicago: University of Chicago Press, 1980).
8
Larry M. Bartels, Presidential Primaries and the Dynamics of Public Choice (Princeton: Princeton
University Press, 1988).
9
For examples, see Aldrich, Before the Convention; Bartels, Presidential Primaries; Samuel L. Popkin, The Reasoning Voter, 2nd ed. (Chicago: University of Chicago Press, 1994).
tain an office more important than the one he now seeks or is holding.10 Of
course, calculating politicians do not blindly seek higher office at every opportunity. Ambition is progressive in the sense that politicians would accept higher
office were it offered to them without risk.11 It is the political opportunity
structure along with an individuals risk profile that partly governs moves up
the political ladder.
The Senate appears to be a natural stepping-stone in Schlesingers view. It
is commonly assumed, for example, that nearly all senators have progressive
ambition for the presidency and place a uniformly high value on holding the
office.12 Unfortunately, there are real risks associated with running for president, especially losing ones current office in the pursuit and ending up out of
government entirely. The political opportunity structure interacts with ambition in this way. Senators are better able to take advantage of the structure than
are other office holders because of their six-year terms. Schlesinger also shows
that senators careers are the most orderly in terms of flowing from office to
office up the hierarchy. And the Senate is seen as higher on the ladder of offices
than are governorships.13 In short, ambition theory suggests that the Senate is
the ideal manifest office for seeking the presidency. Before exploring why
this might not always be true, two caveats are in order.
Historical Milestones
There are two events that might have changed the success rates of senators
seeking the presidency. The first was the 17th Amendment to the Constitution,
which in 1913 provided for the direct election of U.S. senators. Direct election
changed the Senate, making senators more responsive to and more representative of their constituencies.14 One might expect that senators became more successful seeking the presidency after the amendment, because they developed
through generational replacement into a more electorally oriented breed.15
There is some empirical support for this hypothesis. After the 1916 election,
senators share of the contender pool rose from 26 percent to 35 percent, and
their presence among nominees grew from 6 percent to 21 percent. Before the
17th Amendment, senators were less likely to pursue presidential bids and less
10
likely to achieve them if they did run. This had a multiplicative effect that doubly hurt their chances.
The second change occurred when the nomination process was altered in
the 1960s and 1970s. Most notably, the direct primary replaced conventions and
caucuses as the method for choosing nominees. Delegate selection methods,
particularly for Democrats, assured that delegates would be more representative of the citizenry than before. These reforms could have affected senators
success rates if senators were uniquely able to take advantage of primaries or
were somehow hindered by the new rules. These reforms culminated in the
McGovern-Fraser Commission recommendations, which encouraged more
democratic selection of delegates to national party conventions. Using the 1972
election as the first in which these reforms defined the nomination process,
there is no evidence that the process affected recruitment from the Senate. In
the era of direct election, the 1972 reforms did not affect the rate at which senators won their parties nominations.
While these historical changes in the presidential selection structure, particularly direct election, had some effect on senators success rates, they are not
responsible for the persistent gap between theoretical expectations and actual
performance. Anecdotes also fail to generate complete accounts for what
seems to be a systematic pattern; they are better explaining deviations from it.
A fuller explanation of candidate performance is required to augment the rich
histories of individual campaigns and candidates.
Hess, The Ultimate Insiders; Lubalin, Presidential Ambition and Senatorial Behavior.
Lyndon Johnson (1960) and Orrin Hatch (2000) were permitted by state law to run for senator
and president simultaneously.
Expectations
Citizens have different expectations of senators and governors. These are of
several types. Governors must steer the executive branch bureaucracies on a
daily basis while senators address the issues on the national agenda, including
such remote affairs as the annual budget process, treaties, and executive branch
appointments about which voters know little. Governors might be advantaged
because they are expected to act as executives, perhaps the ideal role to play
before becoming the nations chief executive. Though senators such as Al Gore
(2000) sometimes have an advantage in the traditional presidential realm of
foreign policy, the conventional view is that being a governor is like being a
president in miniature.20 Senators are required to compromise and negotiate
18
R. Douglas Arnold, The Logic of Congressional Action (Princeton: Princeton University Press,
1990).
19
David Maraniss, A 20-Year Journeys Lessons, Impressions, Washington Post, 11 August 1996.
20
Louis Harris, Why the Odds Are against a Governors Becoming President, Public Opinion
Quarterly 23 (Fall 1959): 361370. Former presidential candidate Ross Perot explained just before the
2000 presidential election that George W. Bush benefited from this. Perot explained that . . . when
youre a senator, you just pass laws; when youre a governor, you have to make things work. When
youre president, you have to make things work. George Bush has had the practical experience and
demonstrated his ability in being an effective governor and getting results and making things work.
Gore has no experience (NBCs Meet the Press, 5 November 2000).
to be successful legislators, skills that do not necessary make good presidential candidates.
Governors are also perceived as outsiders compared to senators, many of
whom have been in Washington long enough to acquire reputations as Beltway
fixtures. Maybe as a result of these differences, survey data show that governors
are better known than senators within their home states.21 Unfortunately, this
degree of familiarity seems to breed contempt: constituents also view governors
less favorably than senators. Further, governors are also less known around the
country than are senators.22 Except on federal issues such as welfare reform and
the recent tobacco industry settlement, governors are seen as less authoritative
on national matters. The Sunday morning talk shows, after all, rely heavily on
the Senate for their supply of guest experts.23
Finally, governors are disadvantaged to the degree that constituents see
senators linked more to national affairs. This happens in part because the media create these links, even if implicitly, in their coverage of statewide elected
officials.24 Though a senator probably has less responsibility for government in
general, he nonetheless spends more time in Washington than does a governor
and is associated with issues beyond state borders. Research has found that voters perceive these differences and that they vote more on the basis of national
issues in Senate elections and more on the basis of state issues in gubernatorial elections.25
Office Structure
The formal office structures of senators and governors differ in two ways. First,
senators have longer terms. Currently governors terms are four years in all but
two states (where they are even shorter) while senators serve for six years at a
stretch. Further, the switch to four-year terms for governors is relatively new;
in 1960 sixteen states still had two-year terms. Many governors also face term
limitations that senators do not. By state law thirty of the fifty governors must
21
Barbara Hinckley, Richard Hofstetter, and John Kessel, Information and the Vote: A Comparative Election Study, American Politics Quarterly 2 (April 1974): 131158; Peverill Squire and Christina
Fastnow, Comparing Gubernatorial and Senatorial Elections, Political Research Quarterly 47 (December 1994): 703720.
22
Harris, Why the Odds are against a Governors Becoming President.
23
Hess, The Ultimate Insiders; Krasno, Challengers, Competition, and Reelection.
24
Charles M. Tidmarch, Lisa J. Hyman, and Jill E. Sorkin, Press Issue Agendas in the 1982 Congressional and Gubernatorial Election Campaigns, Journal of Politics 46 (November 1984): 12261242.
25
This point is not in contention, though the relative importance of state and national issues is controversial. See Lonna Rae Atkeson and Randall W. Partin, Economic and Referendum Voting: A
Comparison of Gubernatorial and Senatorial Elections, American Political Science Review 39 (March
1995): 99107; Lonna Rae Atkeson and Randall W. Partin, Economic and Referendum Voting and
the Problem of Data Choice: A Reply, American Journal of Political Science 42 (July 1998): 1003
1007; Thomas M. Carsey and Gerald C. Wright, Jr., State and National Factors in Gubernatorial and
Senatorial Elections, American Journal of Political Science 42 (July 1998): 9941002.
now limit themselves to just one or two terms. So when a quadrennial presidential election rolls around, many governors terms are either ending simultaneously or they know that they are lame ducks who must leave office before the
next presidential election.
Only one-third of senators terms end when any given presidents does, and
there are no legal limits on how long they may serve. As a result, senators whose
terms are not up may casually pursue a presidential bid without jeopardizing
their status as elected officials. If the campaign does not succeed in the early
prenomination period, one may simply abandon it and return to life in Washington. The experience might even be helpful in testing the presidential waters
again in the future. Many governors do not have this low-investment luxury.26
When their terms end, most have to pursue another office or return to private
life. They must be serious about their presidential campaigns because it is generally all-or-nothing. The degree of investment should be deeper and more uniform for governors than senators, because they have less to lose.
A side effect of serving six-year terms without limit is that senators are
probably in office longer than are governors. Winning three senatorial elections
puts a politician in Washington for eighteen years, though few governors will
remain in office more than four or eight years total by law. So senators should
be older than governors, have been in office longer, and, most importantly,
have a different view of what makes a successful campaign because of their longevity. The issues and strategies that worked in the first campaign years ago
might not appeal to todays voters as well. Though senators probably invoke
similar themes each time they run, mere name recognition, incumbency perquisites, challenger deterrence, and constituent loyalty are what really contribute
to reelection. This works in ones own state, but could fail miserably on the road
when campaigning nationwide for convention delegates. One of the benefits of
the primary system for parties is that it identifies those candidates who can sell
themselves outside of their home states. Governors, who generally have spent
less time in government, are closer to the voters. Perhaps they are a bit
younger too; the themes and tactics they employed to win a few years ago are
probably still potent today. This might explain in part why young governors
like Jimmy Carter (1976), Bill Clinton (1992), and George W. Bush (2000) succeeded while older Senators Lloyd Bentsen (1988), Scoop Jackson (1980), and
Ernest Fritz Hollings (1988) did not.
A final point on office structure concerns the organization of staff. Senators
have a wealth of staff working for them. They have staff to deal with constituents and interest groups, draft legislation, communicate with other members,
26
As of the early 1990s, thirty-four states held gubernatorial elections at the presidential midterm.
Of these, thirty-one have four-year terms so that governors terms do not usually line up with presidents terms. A few other states hold gubernatorial elections in odd-numbered years. So term limits
and lengths matter more for governors than when their terms begin and end. Many senators, but not
many governors, will be midterm during a presidential election and have been in office for at least one
full term.
manage committee business, and write speeches. This decentralized staff system works well for legislators but is too fragmented to convert easily into a
presidential campaign team. Many of the staff are centered around committees,
caucuses, or policies rather than members. Governors are fortunate to have
more pyramidal staff structures that serve them directly. Many of these aides
were campaign staff who made the transition to governing staff after the initial
election. Loyal and centralized, they can easily be converted back into a personal campaign machine that suits a presidential bid. Many governors have
lieutenant governors who may tend to daily administrative tasks while they are
on the presidential campaign trail. In contrast, sitting senators such as Scoop
Jackson (1972) are simply blamed for missing roll call votes in Washington
when they campaign nationwide.
sought it. This thwarts the aspirations of those in the presidents party who
would like to make a run for higher office. For example, as a strong candidate
for reelection in 1984, President Ronald Reagan faced no real opposition for
the Republican nomination. This had the effect of removing all Republican senators, governors, and other notables from running that year. So the Senates
100 members were not given an equal opportunity to run. Any of fifty-four Republicans in the majority who wished to run probably deferred because of
Reagans presence. Thus, the operative pool of current senators who could
credibly run for president in 1984 was at most forty-six, the number of Democrats at the time. As a result, nine Democrats became contenders, six of them
with senatorial experience. In general, the greater the number of offices held
by the presidents party, the smaller the pool of candidates available to run for
the opposing partys nomination. From 1960 to 1996, the size of the outparty
poolthose senators and governors not from the presidents partyaverages
around seventy-seven individuals. This pool could be reduced further by eliminating those who do not meet the legal requirements (age 35, natural born citizen, and fourteen-year resident) or common demographic requirements such
as being male, white, and Christian.32
TABLE 2
Backgrounds of Presidential Candidates (19601996)
Last Office Held
Contenders
(%)
Nominees
(%)
Winners
(%)
NSRa
VSRb
President
Vice-President
Governor
Senator
Representative
Activist/Celebrity
Federal Appointee
Business Person
Mayor
Sitting Senators
Former Senators
Never in Senate
5.0
6.5
23.0
36.7
10.8
9.4
6.5
1.4
.7
33.8
10.8
57.6
28.6
28.6
19.1
23.8
0
0
0
0
0
23.8
23.8
52.4
40.0
30.0
20.0
10.0
0
0
0
0
0
10.0
30.0
60.0
.86
.67
.13
.10
0
0
0
0
0
.11
.42
.14
.57
.33
.06
.02
0
0
0
0
0
.02
.25
.08
a
b
I have recorded the most recent office or other position held by each of
the presidential contenders as was done in Table 1 by Lubalin and Peabody,
Ornstein, and Rohde. I repeat their analysis in Table 2 using a slightly different
coding of offices held. I include categories for business people such as Steve
Forbes (1996) and activists/celebrities such as Jesse Jackson (1988) and Pat Buchanan (1992 and 1996). As one might expect, senators make up the largest
grouping at more than one-third of all contenders. Governors are not far behind at 23 percent. Aside from House members, no other office comprises more
than a tenth of the total pool.
In addition to the many contenders whose most recent position was in the
Senate, eight other candidates had been in the Senate at some point in their
past; all eight were later chosen president or vice president. To account for all
possible paths from the Senate to the White House, I coded whether they had
ever been in the Senate and, for current senators, whether the ends of their
terms coincided with the presidential election or not.
I rely on two rough measures of success. The nomination success rate
(NSR) is computed as before. The final column also shows the victory success
rate (VSR), which is the ratio of general election winners to contenders. Aside
from those candidates with backgrounds in the House, bureaucracy, or private
sector who were never nominated by their parties, senators have the lowest
success rates, which most importantly are below those of governors. Governors
do 30 percent better at winning nomination (.13/.10) and three times as well at
winning the general election.
The lower half of the table considers the candidates in terms of Senate experience alone to examine how current and former senators differ. This part of
the table shows how senatorial experience anywhere in ones background affects success. The percentage of senators in the pool now grows to 43 percent,
since any contender who ever served in the Senate is counted. Remarkable here
is that only about one in ten candidates was a former senator when running for
president, yet these few contenders performed best, winning nomination almost half of the time and election a quarter of the time. In contrast, current
senators and those without Senate experience do poorly. One would suppose
that current senators invest less in their campaigns than do former senators, as
the theory offered here expects. This finding is especially striking because those
without much relevant experiencerepresentatives, celebrities, and bureaucratswin the general election four times as often as current senators do.
These data clarify the role of the Senate for us. It is only those candidates
with current Senate attachments who underperform as presidential contenders.
Former senators do as theory would expect, but their success is often due to
running from the vice-presidency rather than any direct advantage provided by
their time in the Senate.
This discussion recalls the basic causal argument underlying studies of presidential selection: background experiences shape candidate success. Though a
full account of the factors that affect candidate performance is beyond the
scope of this article, it is useful to consider briefly whether senators differ from
other contenders on some relevant dimensions. Success, of course, may be measured in many ways. Above I measured it using the NSR and VSR. Here I expand the operational definition of success to several continuous measures as
wellpercentage of primary votes for a candidate, percentage of the party convention vote for a candidate, and number of convention delegates won.
I begin with simple demographic data for the 139 contenders in my sample:
their ages and whether they meet the traditional presidential pattern of being
white, male, and Christian. Jews, blacks, and women are thought to be disadvantaged in the process, as are older candidates.34 Because candidates must win
a majority in the Electoral College to be elected, geographic considerations are
important too. I determined whether candidates were from the South or not
and how many electoral votes their home states were allotted. At least in the
post-civil rights era, being from the South is an advantage, evidenced by the
fact that five of the seven presidents elected after JFK were southerners. Being
the favorite son from a state that carries a lot of electoral clout is beneficial too;
all but three of the twenty-two major party presidential nominees since 1960
have won their home states. Finally, two variables capture the visibility or viability of candidates. The first is whether a contender ran before or not, assuming that
an initial failure may lead to success later as it did for Reagan, Nixon, and the
senior George Bush. Second is a measure of the candidates ideological positions. One might expect that those associated with extreme policy proposals are
disadvantaged in the general election, though they may be favored during the
nomination when they seek approval from party activists. I use the absolute value
of NOMINATE scores to assess ideological extremity, which exists for most of
34
TABLE 3
Characteristics of Presidential Candidates (19601996)
Background
Last Office Held
Senator
Governor
Other
Senate Background
Sitting Senator
Former Senator
Sitting & Former
Never a Senator
Electoral
Votes
Age
From
South (%)
Ideology
Extremity
Minority
(%)
Previously
Ran (%)
13.1
23.3
21.5
56.0
54.1
55.1
23.5
34.4
33.9
.359
.361
.324
7.8
18.8
17.9
43.1
40.6
46.4
13.6
16.8
14.3
22.2
56.1
57.1
56.3
54.3
25.5
16.7
23.7
35.0
.350
.403
.361
.328
8.5
0
6.8
20.0
42.6
66.7
47.5
41.3
Note: Ideology measures only exist for former presidents and members of Congress (71 percent
of contenders).
the contenders. (See Appendix.) This variable gives one a rough sense of the
ideological appeal of the candidate to primary and general election voters. Details about the operations of all of the variables may be found in the Appendix.
None of the variables is ideal, nor are they an exhaustive accounting of the
factors that determine presidential candidate performance. While not tapping directly the peculiarities of being in the Senate, each gets at investment differences
among candidates indirectly by assessing their chances for success in much the
same way that donors, activists, voters, or candidates themselves might.
Analysis of Contenders
Table 3 presents mean values for six of the variables by last office held and
Senate background of the 139 presidential contenders. The first data column
reports the mean number of electoral votes for the states from which candidates
run. There are two notable findings here. First, as one would expect, contenders
are disproportionately from large-population states. The average state has 10.8
electoral votes but the typical contender brings 18.8 votes to the campaign (p
.001). More importantly, senators have the lowest number of electoral votes of
the seven types of contenders in the data set. On average they bring to the election ten fewer electoral votes than governors, even though they share the same
constituencies. Though there are counterexamples, senators running for president such as Bob Dole, George McGovern, and Ed Muskie are often from small
states while governors like Jerry Brown, George W. Bush, Ronald Reagan, and
Nelson Rockefeller hail from larger states. This finding is consistent with the
candidate pool explanation provided above. Because governors are a heterogeneous bunch, only the best will make serious runs for the presidency, and they
tend to be from the largest states with the most professional governments. Senators are more homogeneous, so state size matters less in their political calcula-
tions. Since electoral votes and general election victories are correlated (r
.22), the lack of diversity hurts senators as a group.
Senators are also older than governors by a couple of years. While not universal, there are plenty of young governors such as Clinton and George Wallace
and older senators such as Alan Cranston and Dole who have run. This modest
difference is at least consistent with the office structure argument offered
above. Senators have been in office longer because of their six-year terms and
because senators tend to hold more lower offices than governors on their way
up the political ladder.35 The data are also surprising in terms of overrepresentation of the South. Southern states comprise only 22 percent of the Union and
thus of current senators and governors. Even in terms of the Electoral College,
southern states were just 24 percent of the total in 1960 and rose to only 27
percent in 1996. However, the Souths influence grows the further one moves
along the path from candidacy (30 percent) to nomination (33 percent) to victory (50 percent). The fact that the South plays a disproportionately large role
in national politics is not surprising.36 A new implication is that senators low
success rates as contenders stem in part from their nonsouthern bases. If the
U.S. electoral system and voters preferences favor candidates like Johnson,
Carter, and Clinton from below the Mason-Dixon line, then the many senators
who choose to run for president despite being from somewhere else have a
clear liability.
These findings should not imply that senators are always disadvantaged by
their backgrounds. They do not appear more ideologically extreme than other
contenders, for example. In terms of race, religion, and gender, senators are
actually advantaged because they are more likely to fit the traditional pattern
than are other candidates. Only 8 percent are not white, male, or Christian, while
nearly one-fifth of governors and other contenders do not conform with these
historical precedents. Finally, senators are no more reluctant than governors
to give a presidential campaign another shot following a failure.
In summary, this simple analysis has shown at least four ways in which contenders from the Senate differ from governors and other contenders. Senators
are often hindered because they are older, hail from states with fewer electoral
votes, and are less likely to be from the South than are other candidates. To
their advantage, they are more likely to fit the traditional white, male, and
Christian description of American presidents, though the frequency of these
characteristics varies widely across contender types.
To be certain that these inferences are correct, we need to consider more
carefully what it means to be a candidate from the Senate. A contender could
be a current senator, former senator, or without Senate experience when she
or he pursues the presidency. Table 2 showed that former senators performed
35
TABLE 4
Success Rates and Senate Experience (19601996)
Senate Status
Sitting Senators:
Term Up
Sitting Senators:
Term Not Up
Former Senators
Never in Senate
All
Primary Vote
Final
Convention Vote
Nominated
Elected
15
4.1
8.6
6.7
32
12
80
139
9.9
20.5
14.4
12.8
10.1
35.2
13.8
14.2
12.5
41.7
13.8
15.1
3.1
25.0
7.5
7.2
better than expected, while current members and those who never served in
the Senate seldom won nomination and were even less likely to win the election. The bottom half of Table 3 shows mean values of each of the six variables
by Senate background to explore whether differences in success are attributable to other candidate characteristics.
We now find that the differences between senators and other contenders
are really due to differences between sitting senators and others. Former senators do not face as large an Electoral College shortfall, nor, surprisingly, are
they as old as current senators. They benefit from having the right demographic
profiles and experience from previous presidential campaigns. Their only liabilities appear to be having less connection to the South and more ideological diversity than sitting senators.
rent senators according to how much they must commit to a presidential campaign. Senators whose terms end in the presidential election year must often
choose to focus on either the presidential or senatorial campaign, while midterm senators are free to pursue presidential nominations without jeopardizing
their seats.
Clearly the most successful contenders are former senators. Since eight of
twelve were actually running for the vice presidency or presidency when observed, this is not as surprising as it might seem at first. Sitting senators are
much less successful than former senators according to all four criteria. More
importantly, sitting senators also appear to be less successful than those who
were never in the Senate. This implies that the key variable is not Senate experience per se, but being in the Senate while running for president. It does not
seem that legislators necessarily have a more difficult time than other contenders because of the positions they take on roll calls, the negotiating and compromising they do with other members, or the lack of credit they can take for government accomplishments. This result challenges the authority and the issues
and to a lesser degree the expectations explanations offered above. Sitting
members of the Senate, only two of whom ever were elected president, do the
most poorly. Despite his eventual defeat, in retrospect it seems wise of Bob
Dole to have resigned his Senate seat in the summer of 1996 to pursue of campaign full time.
A final point should be made about current senators. Contrary to what the
investment idea implies, senators whose terms are up do a bit worse than those
who are in the middle of their terms. I expected that these candidates would
commit more seriously to the campaign than those who are entertaining the
ideal without real risk. But there is another explanation that still fits with the
ideas of investment and progressive ambition. Senators whose terms expire in
the year in which they want to run for president must decide early in the election which office they will devote themselves to full-time. Senator Phil Gramm
(1996) was one of many presidential candidates whose term happened to expire
in the year he was running. He quickly realized that his presidential campaign
was failing and withdrew before the New Hampshire primary occurred so that
he could focus on winning reelection to the Senate. Today he remains a senator
with the opportunity to run for president again. Senators Dole (1980), Hollings
(1984), and Bentsen (1988) made similar moves by choosing not to invest heavily in presidential bids at the time their Senate seats could have been taken
away. In contrast, senators who are midterm when the presidential election
rolls around may campaign aggressively without fear of losing their place in
Congress. Though not all candidacies fit the pattern, John Kennedy (1960),
Robert Kennedy (1968), and George McGovern (1972) are just three of the
many senators who had the luxury of heavy investment without much riskan
ideal situation.
Conclusion
Contrary to the conventional wisdom found in newspaper stories and academic
journals, senators have been poorer presidential candidates than one might expect. Many of them have run for president, but surprisingly few have earned
their parties nominations and even fewer have won the general election. Former senators do well, though much of their success is attributed to being selected as vice-presidential running mates. The group of presidential candidates
most analogous to senatorsgovernorsmake up the second largest share of
the contender pool and perform better than senators when they run. Though
senators are probably at least as likely as governors to harbor progressive ambition for the presidency, I argue that their differing success rates are due in part
to deeper campaign investment by the typical governor. Governors also benefit
from more careful self-selection, with those from larger states, especially in the
South, more likely to run. It seems evident that senators, especially sitting senators, have not done as well at the nomination and general election stages as they
should have, given what we know. The Senate has seldom been the presidential
incubator or nursery it ought to be given the ambition, visibility, resources, and
records of both current and former members of the institution.
Of the four general explanations offered for the gap between expectations
and performance, some have received more support than others. The authority and the issues and expectations arguments seem the least potent. There
are at least two reasons to be skeptical of these. First, all members of Congress
do not perform equally well. Representatives and senators do have similar success rates in the long run, but members of the House differ significantly from
senators on several predictors of success. Cabinet officials, business people, activists, and mayors have done about as well too, so there does not appear to be
a liability unique to legislators. Representatives should be less successful than
senators, since their constituencies are smaller. They serve in a chamber of 435
rather than 100 and have generally spent less time climbing the hierarchy of
offices to get where they are. Second, current senators are less successful than
former senators. If there is a congressional aura that hurts legislators when they
run for president, it does not extend to members who have left Capitol Hill.
I am more optimistic about the office structure and candidate pool arguments. They have clear connections to the investment and ambition theory
and garner tentative support in the rudimentary analysis done here. They explain why the pools of senators and governors are not quite the sizes that we
assumed if former office-holders are included. Both arguments also rely on notions of supply and demand and strategic decision making by elites, two crucial elements of any study of presidential selection. Mysteries about ascension
of particular candidates to the White House remain of course, but revealing
new facts and proposing explanations for them that are tied to theory have
proved useful.*
* An earlier version of this paper was presented at the 1999 annual meeting of the Midwest Political
Science Association, Chicago. Thanks go to Cecil Eubanks, DuBose Kapeluck, and especially Phil
Appendix
Operational Definitions of Variables:
Electoral Votes: Number of electoral votes for state from which the candidate hails.
Age: Age on general election day of the year in which candidate runs.
South: Candidate from one of the eleven states in the Confederacy.
Ideological Extremity: Absolute value of NOMINATE scores computed by Keith
Poole.37
Minority: Candidate was non-Christian (that is, Jewish), noncaucasian, or female.
Previously Ran: Met the definition of a contender at least once in the past, even if before 1960.
Percent of Primary Vote: Percentage all presidential primary votes earned by the candidate.
Percent of Convention Vote: Percentage of nomination votes on final party convention ballot.
Data Sources:
Michael Barone and Grant Ujifusa, Almanac of American Politics (Washington, DC:
National Journal, various years); Book of the States (Lexington, KY: Council of State
Governments, various years); CQs Guide to U.S. Elections, 3rd ed. (Washington, DC:
Congressional Quarterly, 1994); Congressional Biographical Directory, Biographical
Directory of the United States Congress, 1774Present (http://bioguide.congress.gov),
May 2000; James T. Havel, U.S. Presidential Candidates and the Elections (New York:
Macmillan, 1996); Presidential Elections, 17891992 (Washington, DC: Congressional
Quarterly, 1995); John Runyon, Jennefer Verdini, and Sally Runyon, eds., Source Book
of American Presidential Campaign and Election Statistics 19481968 (New York: Frederick Ungar, 1971); Harold W. Stanley and Richard G. Niemi, Vital Statistics on American Politics (Washington, DC: Congressional Quarterly, various years).
Paolino for their comments and Kimberly Allen, DuBose Kapeluck, and Matt Savrock for their research assistance.
37
Keith T. Poole, Recovering a Basic Space from a Set of Issue Scales, American Journal of Political Science 42 (July 1998): 954993. NOMINATE is a scaling method based on analysis of roll call
votes taken by members of Congress. For presidents who did not serve in Congress, it is based on their
announced positions on bills before the legislature.