The Legislative Success of “Giant Killers”
Sean M. Theriault, The University of Texas at Austin
Jared J. Hrebenar, Stanford University
Isabel Reyna, The University of Texas at Austin
June 2023
Abstract
Since the early 1970s, fewer than 10 percent of all those who run against sitting members of
Congress win. In this paper, we examine the legislative effectiveness of those rare challengers
who knock off incumbents, which we name, “Giant Killers.” We find that they have greater than
expected legislative success. Either because they are simply stellar politicians (as one might
expect from challengers who defeat incumbents) or because their party leadership celebrates
their victories by prioritizing their legislation or giving them plum committee assignments, we
argue that being a Giant Killer is an important component in understanding legislators’ early
successes in Congress. Over the first part of their careers in the House, we find that these Giant
Killers have around 20 percent higher legislative effectiveness scores than their counterparts;
although Senate Giant Killers do not enjoy as large a bonus, it is still real and statistically
significant.
Center for Effective Lawmaking Working Paper 2023-05
2
Sandy Adams (FL-24) and Rich Nugent (FL-11) both first won election to the U.S.
House of Representatives in 2010. On paper, they were nearly indistinguishable: two 50-
something Midwestern transplants to Florida with military experience who parlayed their law
enforcement backgrounds as part of that year’s Tea Party electoral wave. Both had about a
decade of experience in elected office, and both won their elections by more than 20 points.
Indeed, one could reasonably expect them to have similar legislative careers when they
arrived in the nation’s capital as first-term members of the new House majority party. Upon the
beginning of their respective journeys through the marbled halls of Washington, the arcs of their
political careers quickly diverged: Adams proved herself an industrious and effective legislator,
sponsoring the hotly debated reauthorization of the Violence Against Women Act and guiding it
to final passage in the House. She sponsored numerous other pieces of substantive legislation,
never missed a vote, and beat expectations to find a place in the upper half of House Republicans
in terms of legislative effectiveness that congress. Nugent, on the other hand, saw each piece of
legislation that he sponsored die before receiving any action in committee, missed votes, and
ended his first term in Congress with a dismal effectiveness rating.
Admittedly, any number of explanations could be offered to explain this divergence in
lawmaking prowess. From differing personal experiences, gender, staff expertise, leadership
styles, or sheer luck, certain members of Congress simply succeed where others struggle. One
under-analyzed explanation for why these two particular political stories diverged might have to
go back to election day, though focusing a bit less on Adams and Nugent and a bit more on the
names opposite them on the ballot. While Nugent entered Congress the same way most members
do, by waiting for members to retire and running to replace them, Adams entered Congress by
defeating an entrenched incumbent. Because of whom they beat, we name the victors like
3
Adams, “Giant Killers.” While admittedly not all incumbents are giants, the advantages that they
accrue simply by being in office makes their reelection odds much better than those running
against them. Because of the well-known and studied incumbency advantages, it is not surprising
that so few incumbents lose (Jacobson and Carson 2020, Herrnson 2016). What is not well
known, and should not be surprising, is that the very few challengers that beat incumbents are
exceptional, at least in the electoral arena.
In this paper, we argue that they are not only exception in the electoral politics, but also
in the legislative arena. We make this argument in four parts. First, we precisely define who
qualifies as “Giant Killers” and then describe across time their proportion in Congress before
explaining the characteristics that might lead to them having more successful legislative careers.
Second, we determine the precise bonus giant killers have in legislative effectiveness through
bivariate and multivariate analysis in both chambers. Third, we further analyze Senate data to
give our arguments even more validity. Finally, in the conclusion, we suggest multiple paths for
future research.
I. Giant Killers
Theriault and Edwards (2020, 98) assert that “perhaps the best truism of congressional
politics is that incumbents win.” Indeed, the incumbency reelection rate hovers around 95
percent in the House and, though with a bit bigger deviation, around 85 percent in the Senate.
Jacobson and Carson (2020) summarize that the incumbency advantage has multiple sources
including the institutional design of Congress (Mayhew 1974), the changes in voting behavior
4
(Ferejohn 1977), the services members provide to their constituents (Fiorina 1977), their greater
access to campaign contributions (Jacobson 1980), and then, because of these sources, their
ability to scare of challengers (Cox and Katz 1996). Indeed, incumbency is the most important
determinant in understanding who wins congressional elections. And yet, each cycle some lose.
A. Defining Giant Killers
We define any representative or senator who arrived in his or her respective chamber by
beating another incumbent as a “Giant Killer.” This narrow definition introduces a number of
tough cases for which we want to be explicit. First, we include challengers who defeat appointed
senators as Giant Killers.
1
Second, some House members arrived in the chamber after launching
a campaign against an opposite-party incumbent, only to see that incumbent defeated in the
primary. Going on to win that election would not classify that member as a Giant Killer, despite
the fact that the incumbent member was seeking reelection. Third, those members who win their
seats over other incumbents (typically in “redistricting fights”) are not classified as Giant Killers
because they enjoyed the same incumbency advantages as their opponent. While their opponent
may have been a giant, they, too, were a giant, suggesting a status-equal election contest not
unlike two challengers competing for an open seat.
1
We exclude that label from that part of their career prior to when the other senators who won
during the same election are seated. For example, in 2020, Mark Kelly defeated Martha McSally
and served for a few weeks in the 116
th
Congress; he even took a few votes. But his “Giant
Killer” status does not kick in until the beginning of the 117
th
Congress, which was when the
other senators newly elected on election day 2020 took their seats.
5
We offer one additional tweak. Giant Killer status is not necessarily determined at a
members’ first-ever term in Congress. Senators or representatives who retired, were defeated, or
otherwise left the chamber, but then returned to the chamber by defeating an incumbent, are
coded as Giant Killers for the congresses after their return (but, obviously, not before unless they
obtained that seat by also defeating an incumbent then). We viewed this “status” as something
that could only be gained, never lost: for instance, in Barry Goldwater and Hubert Humphrey’s
returns to the Senate after their stints in the national spotlight, both retained their Giant Killer
identity that they earned in their first elections to the Senate, despite neither defeating
incumbents in their second go-round.
2
We further subset Giant Killers into two categories depending upon if they defeated
incumbents during a primary or general election. The dynamics in these elections are likely to be
quite distinct. Rarely would Primary Giant Killers enjoy the support of their party in the primary
elections, whereas the General Election Giant Killers would almost always have significant party
backing as they try to pull off the rare incumbent defeat. By separating General Election Giant
Killers from Primary Giant Killers, we can discern the differences in legislative effectiveness not
only between Giant Killer and non-Giant Killers, but between the subsets of Giant Killers
themselves, which as we explain below will help tease out potential explanations for Giant
Killers’ greater legislative success.
2
We should add that none of these tough cases change the overall results we report in this paper.
Had we made the opposite call on each of these decisions, the thrust of the results would not
change. It could have to do the with the robustness of our results or the fact that so few members
actually fall into these tough cases.
6
B. The Frequency of Giant Killers Across Time
In aggregate from the 93
rd
through the 115
th
Congress, Giant Killers comprised 28
percent of the House, though the number of Giant Killers has gone down over time. Not
unsurprisingly, the two largest new class of Giant Killers entered the House in 1974 (Watergate
Babies) and 2010 (Tea Partiers). The broad sweep of the last 50 years in the House shows that
most representatives are not Giant Killers; furthermore, General Election Giant Killers are far
more common than their Primary Giant Killer counterparts.
7
Figure 1: The Number and Type of Giant Killers in the House of Representatives, 93
rd
to 115
th
Congresses (1973-2018).
300
288
302
306
302
309
307
312
318
319
323
317
325
328
339
342
341
340
331
311
310
312
330
33
30
23
23
20
20
20
22
22
22
14
19
21
20
18
18
19
20
20
18
18
24
23
4
6
3
4
6
3
3
2
1
1
11
3
1
1
0
4
2
2
2
3
8
4
5
91
73
98
87
82
83
93
98
97
84
72
65
75
82
78
71
69
61
72
56
84
84
79
12
42
13
18
32
24
16
7
6
15
20
35
22
6
7
4
8
24
20
55
24
13
5
93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115
Not Giant Killers Returning Primary Giant Killers
New Primary Giant Killers Returning General Election Giant Killers
New General Election Giant Killers
8
Giant Killers are a relatively larger proportion of the Senate than they are of the House,
though their numbers have more recently shrunk (see figure 2). Whereas they constituted nearly
50 percent of the Senate in the beginning of our dataset, their numbers have been almost halved
in the last few congresses. The largest class of new Giant Killers entered the Senate in 1981, the
same year that Reagan entered the White House and brought with him a new Republican-
majority Senate. The number of Primary Giant Killers has experienced the largest decline. In the
beginning of our data set, they sat in 8 or 9 seats in the Senate; by the end, Mike Lee (R-UT)
constituted a conference of just one.
9
Figure 2: The Number and Type of Giant Killers in the Senate, 93
rd
to 115
th
Congresses (1973-
2018).
56
56
59
54
54
56
59
60
59
61
62
63
64
69
67
63
68
70
71
66
74
73
75
7
6
8
7
7
7
7
6
6
6
5
6
5
3
3
3
3
3
2
1
2
1
1
1
2
0
2
0
0
0 0
0
0
1
0
1
0
0
1
0
0
0
1
0
0
0
32
33
27
31
29
36
32
28
32
33
28
30
29
27
26
30
29
23
28
31
27
21
26
5
3
8
7
10
2
3
7
4
1
5
2
1
3
5
3
1
6
5
2
1
5
2
93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115
Not Giant Killers Returning Primary Giant Killers
New Primary Giant Killers Returning General Election Giant Killers
New General Election Giant Killers
10
C. Why Giant Killers May Have Greater Legislative Success
When new members are sworn into office, they all take the same oath and are generally
treated in the same freshman kind of way. In the House, seniority among those sworn in on the
same day is determined by lot (Kellermann and Shepsle 2009). The Senate procedure, while not
as random, still does not prioritize Giant Killers in any way.
3
Nevertheless, we think that Giant
Killers might be more effective for two different reasons: one having to do with who they are
and the other having to do with who they beat.
First, by virtue of beating an incumbent, Giant Killers may simply be exceptionally
talented politicians. Just as it is not easy to beat an incumbent, it is not easy to pass legislation,
and the skills that helped an individual do the former may prove useful at achieving the latter.
While getting the details right in legislative drafting is not the same thing as developing policy
papers, and while assembling an enacting legislative coalition is not the same thing as getting
campaign contributions, the organizational and interpersonal skills on the campaign trail my not
be that unrelated to those that are required for a successful legislative record.
It should be noted that being a Giant Killer may also be an indication or either a more
carefree approach to electoral politics or an utter disregard for the political science literature.
Both by anecdote and study (Jacobson and Kernell 1983; Maisel and Stone 1997; Carson et al
2011), the most strategic politicians are most likely to wait for an open seat to launch their
3
For those taking the oath on the same day, seniority is determined by “former service in order
as senator, vice president, House member, cabinet secretary, governor, and then by state
population.” (https://www.senate.gov/artandhistory/history/resources/pdf/chronlist.pdf; accessed
on 14 January 2020).
11
congressional campaigns. While individual personal circumstances (e.g., reaching the end of a
term limit for state legislator or a free attempt at Congress while serving in an office with a
different reelection date) or strategic calculations (e.g., a vulnerable incumbent either because of
malfeasance or incompetence) may compel even risk-averse ambitious politicians to throw their
hats in the ring against an incumbent, few challengers are expected to win their races at the time
they announce their candidacies. Because the odds are so low, most strategic politicians opt for
the more likely path to Congress – winning an open seat.
Second, because most Giant Killers earn that distinction by capturing a seat previously
held by the other party, perhaps their party leadership treats them differently when they arrive in
the hallowed halls of the Capitol. In fact, it could be that the promises that leadership made to a
successful businesswoman or a popular sportscaster that compelled them to pause their
professional careers to try their hand at politics. Once they take their seats in Congress,
leadership may reward them with better committee assignments or may prioritize their legislative
goals in hopes of making a vulnerable first-term member safer.
Distinguished Primary Giant Killers from General Election Giant Killers may help parse
any legislative bonus that Giant Killers might enjoy. For however hard it is to defeat an
incumbent in a general election, it is even harder in a primary, when candidates typically cannot
even rely upon their party to assist them, and in some instances have their party vigorously
oppose them. As a result of climbing an even steeper mountain, these Primary Giant Killers may
be truly exceptional politicians. But, counteracting that exceptionalism, once Primary Giant
Killers show up in Congress, they may face lingering hard feelings from the party conference or
caucus that used to contain their primary opponent, which may make passing their bills even
more difficult. In essence, then, the existence of Primary Giant Killers may help parse out why it
12
is that Giant Killers enjoy greater legislative success. And, the distinction between the House,
where members have shorter time horizons and more anonymous existences (Asher 1973), and
the Senate, where interpersonal relationships matter more (Matthews 1960, Theriault 2013)
II. The Legislative Effectiveness of Giant Killers
The 1990’s professional wrestler, Ric Flair, made famous the concept we explore in this
paper: “To be the man, you have to beat the man.”
4
While the data in the previous section
suggests that most members of Congress just wait until the “man” or “woman” retires and then
runs for his or her seat, sufficient members do not wait until a retirement for us to examine if
they have a fundamentally different experience within Congress than those who gain their seats
through open-seat races. Our dependent variable is the members’ legislative effectiveness scores
(LES), which are derived from the success and importance of the legislation members sponsor
(Volden and Wiseman 2014, 2018). The scores have a mean of 1 in both chambers every
congress. They range from 0 (no introduced bills that get any legislative traction) to 18.7 in the
House (Charlie Rangel in the 110
th
Congress) and 10.1 in the Senate (Howard Cannon in the 96
th
Congress).
4
Please excuse the gender-exclusive language. Quoted in Zack Pumerantz, October 25, 2011,
“The 100 Best Sports Quotes of All Time,” The Bleacher Report
(https://bleacherreport.com/articles/910238-the-100-best-sports-quotes-of-all-time; accessed on 8
August 2020).
13
At the simplest level, the data show that Giant Killers are marginally more effective than
those members elected in open-seat contests. For all congresses and coding a members’ status for
the duration of their entire careers in Congress, Giant Killers in the House have a mean LES of
1.05 in the House (and a mean LES of 1.01 in the Senat), whereas their counterparts’ mean is
0.98 (0.99) – though small, the difference has a p-value of 0.03 (0.32), which meets the standards
of statistical significance in the House. If we isolate the comparison to members’ first term in
Congress, the difference is greater: Giant Killers in their first term have a mean LES of 0.56
(0.41), while their counterparts’ mean is 0.43 (0.40). At least in the House, Giant Killers in their
first term are 30 percent more effective than those representatives who win open-seat contests.
5
Before testing to see if these distinctions can withstand the comprehensive multivariate
analysis presented in Volden and Wiseman (2014, 2018), we first explore the difference between
Giant Killers and their counterparts across time and over the tenure of their congressional
careers. The across-time analysis has a similar pattern in both chambers. Up until the mid-point
of our analysis, Giant Killers are not that distinct from their counterparts, though starting in the
104
th
Congress (1995-6) in the House the patterns diverge (see figure 3, panel A). The crossover
point happens 3 congresses later in the Senate (see figure 3, panel B). In the House, Giant Killers
did 0.05 better in the first half and 0.08 better in the second half. The difference over time is even
greater in the Senate where Giant Killers did 0.09 worse in the first half, but 0.21 better in the
second half.
6
5
The House difference has a p-value of 0.0001; the Senate’s p-value is 0.41.
6
Neither of the differences in the first half is close to statistical significance; both of the
differences in the second half easily reach it (p-value=0.03 in the House; p=0.007 in the Senate).
14
Figure 3: The Legislative Effectiveness Scores for Giant Killers and Their Counterparts, 93
rd
to
115
th
Congress (1973-2018).
Having isolated the importance of being a Giant Killer in the second half of our study’s
time frame, we now examine the effect of being Giant Killers over the course of their
congressional careers. It could be that the ability to knock off an incumbent is simply an
indication of being a better politician or the euphoria that a Giant Killer experiences leads to an
increase in the beginning of their congressional career that dissipates over time as open-seat
winners gain legislative experience and the importance of being a Giant Killer wanes. Again, the
data show a similar trend in both chambers, though with a different “break-even” point – and,
0
0.2
0.4
0.6
0.8
1
1.2
93 95 97 99 101 103 105 107 109 111 113 115
A. The House of Representatives
Not Giant Killers
Giant Killers
0
0.2
0.4
0.6
0.8
1
1.2
93 95 97 99 101 103 105 107 109 111 113 115
B. The Senate
Not Giant
Killers
Giant
Killers
15
again, the Senate lags the House (see figure 4). The distinction between Giant Killers and their
counterparts in the House is real for the first ten years. In the Senate, it is not until their ninth
congress that the counterparts do as well as the Giant Killers.
16
Figure 4: The Legislative Effectiveness Scores for Giant Killers and Their Counterparts over the
Course of their Careers.
The boost that Giant Killers get in legislative effectiveness appears to be greatest in the
second half of our study’s timeframe and the first half of most members’ congressional careers.
The simple data analysis from this section is also validated by a consistent finding from Volden
and Wiseman (2014, 2018) that members’ effectiveness increases with their seniority.
Furthermore, our analysis shows that the effectiveness is greatest in the point of their careers
where the Giant Killer boost has completed dissipated.
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
1 2 3 4 5 6 7 8 9 10+
A. The House of Representatives
Not Giant Killers
Giant Killers
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
1 2 3 4 5 6 7 8 9 10+
B. The Senate
Not Giant Killers
Giant Killers
17
When both of these findings are taken into consideration, the real effect of being a Giant
Killer is stark. Giant Killers in their first five congresses in the House after the 104
th
Congress
are more than 20 percent more effective than their counterparts. Giant Killer senators in their
first nine congresses after the 108
th
Congress are 17 percent more effective.
III. The Giant Killer Effect while Controlling for Other Variables
To determine if the Giant Killer effect is real, we include its operationalization in the
comprehensive models that Volden and Wiseman (2014, 2018) developed for each chamber. We
first test their model using the data up through the 115
th
Congress (2017-8), though we did make
a couple of minor tweaks to both chambers’ regression models. In the House, we used the
incumbents’ two-party vote rather than their percentage of the vote that they received on election
day.
7
Also, to include those members who were elected in special elections we used an indicator
variable for those members that served the full 2-year term.
8
In the Senate, to include those
senators who were appointed to office we use an indicator variable for those members who were
not appointed, which becomes an interaction term for vote percentage. For both variables,
appointed senators have a value of 0; the 0 for the vote percentage will have no affect because of
7
Because they differ only with the presence of third-party candidates, which we think slightly
distorts the electoral safety measure, the two-party vote and win percentage are highly correlated.
8
Because they are necessarily elected in open seats, we wanted to include even those elected in
special elections.
18
the 0 for the indicator variable.
9
Rather than maintaining consistency across chambers, we
maintain consistency with the sets of independent variables that Volden and Wiseman (2014,
2018) used, which, differed slightly.
To operationalize the Giant Killer effect appropriately, we include several variables in
both chambers’ regressions. First, an indicator variable that takes a value of “1” if the member is
a Giant Killer and in the first part of their congressional career (to maintain consistency with the
previous section, the cutoff in the House is 5 terms and in the Senate is 8 terms). Second, to
account for the across-time change in the data analysis in the previous section, we include an
indicator variable if the congress for the Legislative Effectiveness Scores is in the second part of
the data analysis (beginning with the 104
th
Congress in the House and 107
th
in the Senate). Third,
an interaction variable of these first two indicator variables. If the Giant Killer effect from the
previous section is robust to multivariate analysis, this interaction term should be positive. We
now discuss the results by chamber.
A. The Giant Killer Effect in Multivariate Analysis: The House
We closely match the Volden and Wiseman (2014) House results (see column A of table
1), which provides confidence at two different levels. First, we feel that it demonstrates that we
have properly simulated their results. Second, it suggests that the relations that they thought were
important are still relevant on data up through the 115
th
Congress.
9
Because we include appointed senators as incumbents, we thought we should capture the effect
they have on the tests even in the congress to which they were appointed.
19
Table 1: The Legislative Success of Giant Killers in the House of Representatives.
(A)
(B)
Volden-Wiseman Results
With Giant Killer Variables
Seniority
0.059
***
0.060
***
(0.01)
(0.01)
State Legislative Experience
-0.112
**
-0.115
**
(0.07)
(0.07)
State Legislative Experience *
0.520
***
0.543
***
Legislative Prof.
(0.21)
(0.21)
Majority Party
0.440
***
0.411
***
(0.04)
(0.05)
Majority-Party Leadership
0.458
***
0.457
***
(0.15)
(0.15)
Minority-Party Leadership
-0.133
***
-0.125
**
(0.05)
(0.05)
Speaker
-0.677
***
-0.688
***
(0.24)
(0.24)
Committee Chair
3.059
***
3.059
***
(0.23)
(0.23)
Subcommittee Chair
0.764
***
0.770
***
(0.07)
(0.07)
Power Committee
-0.212
***
-0.211
***
(0.05)
(0.05)
Distance from Median
0.009
-0.068
(0.09)
(0.11)
Female
0.078
**
0.060
*
(0.05)
(0.05)
African American
-0.290
***
-0.287
***
(0.07)
(0.07)
Latino
0.000
-0.017
(0.10)
(0.10)
Size of Congressional Delegation
-0.003
**
-0.003
**
(0.00)
(0.00)
Full Term
0.008
-0.010
(0.11)
(0.11)
Win Percentage
0.014
*
0.017
**
(0.01)
(0.01)
Win Percentage
2
0.000
*
0.000
**
(0.00)
(0.00)
Giant Killer (First Five Terms)
-0.049
(0.05)
Post 104th Congress
0.051
(0.05)
Giant Killer (First Five Terms) *
0.231
***
Post 104th Congress
(0.07)
Constant
-0.341
-0.450
(0.37)
(0.37)
20
N=10,080; ***Statistically significant at 0.01; **Statistically significant at 0.05; *Statistically significant at
0.1.
We first test to see if the increased legislative success for Giant Killers that we observed
in the bivariate results persists in the multivariate test (column B). Before the 104
th
Congress,
being a Giant Killers results in a lower LES of 0.049, though that estimate is not statistically
significant. After the 104
th
Congress though Giant Killers have a 0.233 higher LES.
10
This effect
is both substantively and statistically significant. Given that the LES average for members in
their first five terms is only 0.674, members who knock off an incumbent have LES scores one-
third higher than their counterparts.
We can put this effect into a broader context. The regression output suggests that the
boost gained from being a Giant Killer in the first five terms during the latter part of the time
period we examine is almost four times as much as that gained for being a woman, a result that
Volden, Wiseman, and Wittmer (2018) trumpeted with great fanfare. Furthermore, it is worth
almost 4 congresses of seniority, about half as much as being in the majority party, and about
one third of being a subcommittee chair. All of this is to suggest that those challengers who beat
incumbents on their way to taking their seat in the House chamber significantly overperform
their counterparts who won in open seats.
In additional analyses that we performed, we could not uncover a systematically bigger
effect for Primary Giant Killers from General Election Giant Killers. It could be that their even
better political skills equally offset the punishment of beating a caucus or conference friend or
that by the time the member takes the oath of office in the House all that matters is the letter –
10
The 0.233 Giant Killer effect is determined by adding the indicator variable coefficients for
Giant Killers in their first five terms and post 104
th
Congress to the indicator variable.
21
“D” or “R” – behind their name. In a chamber where numbers matter more than personalities,
these results were not altogether that surprising.
B. The Source of the Increased Legislative Success
We gain further insight into the potential sources of this legislative boost by substituting
our dependent variable in the above multivariate analysis for the various stages of the legislative
process included in the calculation of the LES – allowing us to determine if, for instance, Giant
Killers are more effective because they simply introduce more bills or because a higher
proportion of the bills they introduce move along in the legislative process. When we break this
process down, a few noteworthy results emerge, again for our population of post-104th Congress
Giant Killers in their first five terms. Giant Killers, on average, have 0.74 more bills reported out
of committee and 0.49 more bills pass the House.
11
Part of this greater success could be a
consequence of the Giant Killers introducing two more pieces of substantive legislation on
average than their peers,
12
even as they sponsor a similar amount of non-substantive legislation.
13
We now consider the smallest subset of our dataset – Primary Giant Killers – and attempt
to ascertain what differentiates them from their general election counterparts. From the dataset, a
11
Both findings are statistically significant at 0.01.
12
Statistically significant at 0.05.
13
Several other areas were explored to see if Giant Killers displayed a boost along certain stages
of the legislative process compared to non-Giant Killers, or if General Election Giant Killers and
Primary Giant Killers differentiated from one another at any step of the process. None of these
results proved statistically significant.
22
few insights allow us to construct a narrative: we find that Primary Giant Killers sponsor a
notably higher proportion of commemorative bills compared to their general election and non-
Giant Killer counterparts, a differential that only grew after the 104th Congress. This result could
suggest an attempt by Primary Giant Killers to mend fences after presumably upsetting important
interest groups and power players in their district when they defeated the incumbent. In essence,
we posit that the increased proportion of commemorative legislation points to their attempts to
smooth over resentment that they may have caused when they defeated a member of their own
party – honoring individuals in their district, renaming post offices, and generally being seen
“working for the district” as much as possible. These conclusions, of course, are speculative, but
consistent with our deeper dive into the legislative activities of the Giant Killers.
In our consideration of Primary Giant Killers, we also examined a possible exception to
our working theory that they were likely to enter the halls of Congress with some level of
resentment from their colleagues. Though exceedingly rare, on occasion, party leadership wanted
certain incumbents to be defeated, which would then make our Primary Giant Killers not a
resented usurper, but instead a welcomed replacement for a particularly problematic member of
the caucus or conference. Examples of these situations are not always clear, but we examined a
few obvious cases to check our argument. Our population of “party-desired” Primary Giant
Killers include Representative Mo Brooks, who ousted a former Democrat in the primary to take
his seat in the House in 2010, Representative David Trott, who defeated a strong Tea Partier in a
2014 primary that party leadership was worried would lose the seat, and Representatives Hansen
Clarke and Dwight Evans, who ousted incumbents plagued by scandal in the 2010 and 2006
primaries, respectively. We could detect no obvious benefit accrued to this group upon their
arrival in Washington – suggesting that the party leaders did not spend additional resources in
23
supporting the new member, even if they were pleased that they beat the incumbent. On the
contrary, we found that this group had a below average LES, though not in large enough
numbers to constitute a statistically significant result. Perhaps these members thought that by
virtue of defeating a hated member of the caucus they felt that they did not need to deliver as
much to their district. Or, alternatively, most of these members ran in safe seats, which reduced
the incentive for their party leaderships to disproportionately help them upon taking the oath of
office. This dataset is, admittedly, too small to derive definitive findings, but the trend does
provide some insight on the nature of Primary Giant Killers.
C. The Giant Killer Effect in Multivariate Analysis: The Senate
As with the House results, we are able to closely match the Volden and Wiseman (2018)
results in the Senate (see column C of table 2). Again, we test to see if the legislative success
among Giant Killers that we observed in the bivariate results persists in a multivariate test
(column D). Before the 106
th
Congress, being a Giant Killers results in a -0.089 lower LES.
After the 106
th
Congress though Giant Killers have a 0.030 higher LES. Admittedly, the
differences are not huge, but they are big enough and consistent enough that the variables are
jointly significant.
24
Table 2: The Legislative Success of Giant Killers in the Senate
(A)
(B)
(C)
Volden-Wiseman Results
With Giant Killer Variables
With GK Variables by
Election
Majority Party
0.352
***
0.347
***
0.346
***
(0.09)
(0.09)
(0.09)
Committee Chair
1.046
***
1.041
***
1.041
***
(0.11)
(0.12)
(0.11)
Subcommittee Chair
0.199
***
0.204
***
0.202
***
(0.08)
(0.08)
(0.08)
Majority-Party Leadership
0.017
0.009
0.004
(0.16)
(0.16)
(0.16)
Minority-Party Leadership
-0.115
**
-0.121
**
-0.124
**
(0.06)
(0.06)
(0.06)
Power Committee
-0.150
***
-0.151
***
-0.148
***
(0.06)
(0.06)
(0.06)
State Legislative Experience
-0.212
**
-0.214
**
-0.215
**
(0.11)
(0.11)
(0.11)
State Legislative Experience *
0.958
**
0.974
**
0.965
**
Legislative Prof.
(0.47)
(0.48)
(0.48)
Distance from Median
-0.057
-0.073
-0.091
(0.12)
(0.13)
(0.13)
Majority-Party Women
-0.002
-0.025
-0.029
(0.14)
(0.14)
(0.14)
Minority-Party Women
0.243
***
0.226
**
0.227
**
(0.10)
(0.10)
(0.10)
Freshman
-0.259
***
-0.264
***
-0.263
***
(0.05)
(0.05)
(0.05)
Seniority
0.093
***
0.092
***
0.092
***
(0.02)
(0.02)
(0.02)
Seniority
2
-0.004
***
-0.004
***
-0.004
***
(0.00)
(0.00)
(0.00)
African American
-0.102
-0.096
-0.058
(0.11)
(0.11)
(0.11)
Latino
0.125
0.120
0.120
(0.14)
(0.14)
(0.15)
Not Appointed
0.022
0.036
0.035
(0.10)
(0.10)
(0.10)
Win Percentage
0.002
0.002
0.003
(0.00)
(0.00)
(0.00)
Win Percentage
2
0.000
0.000
0.000
(0.00)
(0.00)
(0.00)
Giant Killer
-0.089
*
(First Eight Terms)
(0.06)
General Election GK
-0.053
(First Eight Terms)
(0.07)
Primary GK
-0.252
***
(First Eight Terms)
(0.08)
Post 106th Congress
-0.003
0.000
(0.06)
(0.06)
Giant Killer *
0.122
Post 106th Congress
(0.10)
General Election GK *
0.098
Post 106th Congress
(0.10)
Primary GK *
0.147
*
Post 106th Congress
(0.10)
Constant
0.325
**
0.352
**
0.341
**
(0.15)
(0.15)
(0.15)
N=10,080; ***Statistically significant at 0.01; **Statistically significant at 0.05; *Statistically significant at
0.1.
25
As we did with the House results, we can put these substantive results into a broader
context. Being a Giant Killer in the Senate is worth slightly more than one-third of an additional
congress of experience and about one-seventh of being a subcommittee chair or a woman serving
in the minority party. Again, this effect is not substantively large, but the joint statistical
significance suggests that it is real.
IV. One Final Word about the Senate
We will be honest. Once we saw the results for the two chambers, we were inclined to
simply write this paper about the House and pretend – like many other congressional scholars –
that the Senate does not exist. A deeper dive into the Senate data, though reveals an interesting
pattern that speaks to the distinction between Primary versus General Election Giant Killers. The
House data analysis suggested that both had similar effects on their effectiveness (or, perhaps the
extra ability of pulling the rare primary win was almost equally offset by the rebuke the party
leveled for defeating one of their former caucus or conference members). In the Senate, a
distinction existed, and it rang true to how we thought it would.
The results (see column C from table 2) show both consistency and difference across the
stages of when newly elected members earn their Giant Killer status in the Senate. The
differences between the pre-106
th
and post-106
th
are consistent. Giant Killer status is 0.147
greater in the latter period for Primary Giant Killers and 0.099 greater for General Election Giant
Killers. Where they differ is that Primary Giant Killers actually are 0.199 less effective in the
earlier period and 0.149 less effective in the latter period than General Election Giant Killers.
26
Again, drawing these distinctions does not necessarily show statistical significance in all the
variables, but the set of variables are jointly significant.
Once the punishment suffered by Primary Giant Killers is separated from the reward of
General Election Giant Killers, the substantive effect of that reward increases to 0.044, though
still not nearly as large as the House effect. Being a General Election Giant Killer is worth about
a year in the Senate and about one-fifth as much as being a subcommittee chair or a woman in
the minority party.
The contrast between the House results (not shown) and the Senate results on when the
member earns their Giant Killer status is also consistent with how we think the two chambers
and their members operate. The sore feelings senators may have toward the new kid who shows
up after beating one of their friends is simply greater in a chamber where personal relationships
still matter. In the “People’s House,” personal relationships do not matter as much as merit and
ability – at least according to the tea-leaf reading of our results suggests.
V. Conclusion
This paper bridges one of the oldest and most developed paths of congressional research
– the incumbency advantage – with one of the newest and most exciting paths for future
congressional research – the determinants of legislator success. While incumbents win the vast
majority of the races that they run, when they do not win, the challengers to whom they lose go
on to have more productive legislative careers than those members who gain their office in open-
27
seat contests. We find that this effect is real and pronounced for the first 5 terms in the House.
Afterward, it appears that the Giant Killers bonus disappears.
While the Senate Giant Killer bonus is not as big, it is still statistically significant.
Furthermore, when the election during which they earn the status is taken into consideration, we
find that the bonus only accrues to General Election Giant Killers; Primary Giant Killers have
less legislative success. These results suggest that while they may be even better politicians for
defeating an incumbent in a primary, they suffer a stigma once they join the conference or
caucus to which their vanquished opponent used to be a member.
We see at least two future directions for the research that we present here. First, we could
imagine parsing out more systematically why Giant Killers overperform. Is it that they are
simply better politicians? Or are they given better committee assignments or treated better by
party leadership because of their vaulted or vulnerable electoral status? Second, the findings
from this paper may suggest that those challengers who defeat incumbents constitute a subset of
legislators that are simply different, not unlike those who were elected in a particular election
(Watergate Babies, Tea Party Members, or Nancy Pelosi’s Majority Makers) or who had
particular experiences (the Gingrich Senators).
28
Bibliography
Asher, Herbert B.1973. “The Learning of Legislative Norms.” American Political Science
Review, 67(2): 499-513.
Carson, Jamie L., Michael H. Crespin, Carrie P. Eaves, and Emily Wanless. 2011. “Constituency
Congruence and Candidate Competition in U.S. House Elections.” Legislative Studies
Quarterly 36: 461-482.
Cox, Gary, and Jonathan Katz. 1996. “Why Did the Incumbency Advantage in the U.S. House
Elections Grow?” American Journal of Political Science 40: 478-497.
Ferejohn, John. 1977. "On the Decline of Competition in Congressional Elections." American
Political Science Review 71(3): 166-176.
Fiorina, Morris P. 1977. “The Case of the Vanishing Marginals: The Bureaucracy Did It.”
American Political Science Review 71 (3): 177-181.
Herrnson, Paul S. 2016. Congressional Elections: Campaigning at Home and in Washington, 7
th
edition. CQ Press.
Jacobson, Gary C. 1980. Money in Congressional Elections. New Haven, CT: Yale University
Press.
Jacobson, Gary C., and Jamie L. Carson. 2020. The Politics of Congressional Elections, 10
th
edition. Lanham, MD: Rowman & Littlefield.
Jacobson, Gary C., and Samuel Kernell. 1983. Strategy and Choice in Congressional Elections,
2
nd
edition. New Haven, Connecticut: Yale University Press.
29
Kellermann, Michael, and Kenneth A. Shepsle. 2009. “Congressional Careers, Committee
Assignments, and Seniority Randomization in the U.S. House of Representatives,”
Quarterly Journal of Political Science 4(2): 87-101.
Maisel, Sandy, and Wlater P. Stone. 1997. “Determinants of Candidate Emergence in U.S.
House Elections: An Exploratory Study,” Legislative Studies Quarterly 22(1): 79-96.
Matthews, Donald. 1960. U.S. Senators and Their World. Chapel Hill: University of North
Carolina Press.
Mayhew, David R. 1974. Congress: The Electoral Connection. New Haven, CT: Yale University
Press.
Theriault, Sean M. 2013. The Gingrich Senators: The Roots of Partisan Warfare in Congres.
New York: Oxford University Press.
Theriault, Sean M., and Mickey Edwards. 2020. Congress: The First Branch. New York: Oxford
University Press.
Volden, Craig, and Alan E. Wiseman. 2014. Legislative Effectiveness in the United States
Congress: The Lawmakers. New York: Cambridge University Press.
Volden, Craig, Alan E. Wiseman, and Dana E. Wittmer. 2013. “When Are Women More
Effective Lawmakers Than Men?,” American Journal of Political Science 57(2): 326-
341.
Volden, Craig, and Alan E. Wiseman. 2018. “Legislative Effectiveness in the United States
Senate,” Journal of Politics 80(2): 731-735.
