Running in Someone Else ’ s Shoes : The Electoral Consequences of Running as an Appointed Senator

Over the past century, nearly two hundred times a governor has appointed an individual to fill a vacant Senate seat. This research seeks to understand the electoral fates of these appointed senators. First, I address the question of when and under what conditions an appointed senator will choose to run for reelection to the seat. Then, should they choose to run for that office in the next election, they are in the rare position of being an incumbent who has not previously won an election to that particular office. Although these appointed senators are not on equal footing as other first-term senators, they still provide a unique circumstance worthy of further examination. I find that those appointed senators who had previously held an elected office were more likely to run to maintain the Senate seat. I also find that appointed senators fare slightly worse than other first-term senators did when campaigning for reelection.


Introduction
When a vacancy occurs in the Senate, because of the death or retirement of a sitting senator, the governor of that state is able to appoint an individual to fill that seat until a special election can be held.These appointed senators have the same rights as an elected senator and may serve for several years before the next election occurs.Since the passage of the Seventeenth Amendment and the ensuing direct election of senators, nearly 200 appointed senators have filled such vacancies.This research examines the fates of these appointed senators, asking two main questions.First, under what conditions would we expect an appointed senator to run for reelection to the Senate seat?Second, how do appointed senators fare as compared to first-term senators seeking reelection?Not all appointed senators choose to seek reelection.Some, like Rebecca Felton (D-GA), the first woman to serve in the Senate, are appointed symbolically and plan only to serve until an election for a replacement can be held. 1 Others still are appointed, serve, and seek reelection only to then fail to secure their party's nomination and, therefore, are unable to run in the general election.The first question I address is under what conditions an appointed senator is likely to seek election to the office.Then I compare the electoral fates of those appointed senators who are candidates in the general election to other first-term senators who run for reelection.
Relying on data from Senate elections from 1920 to 2012, I find that the choice to seek election to the Senate for appointed senators can be idiosyncratic and difficult to predict.However, the appointed senators' age and length of service in the Senate prove to be two significant factors in predicting the decision to run.Another factor that makes an appointed senator significantly more likely to run for the seat is if that individual has ever previously held an elective office.Those who have previous experience running and winning a political campaign are more likely to run for the Senate seat.A significant portion of the appointed senators chose to retire immediately and not campaign for office; in addition, a small proportion of them lose their party's nomination contest.For those who do run in the general election, at first glance it seems that their electoral fortunes do not differ much from their counterparts who are completing their first-term in the Senate.However, an Ordinary Least Squares regression shows that appointed senators in fact fare worse.I also include a logistic regression to predict the likelihood of winning the election.Again, appointed senators are less likely to win than other first-term senators who were elected to the Senate six years prior.

Filling Senate Vacancies
Vacancies in the Senate can arise for a variety of reasons, the most obvious being the death of a sitting senator.However, they also frequently occur because a senator chooses to resign from office during the middle of a six-year term.Frequently a number of vacancies occur at the beginning of a new president's first term as he often appoints a number of senators to various cabinet posts.For example, in 2009, there were four Senate vacancies created as both President Obama and Vice President Biden vacated their Senate seats, along with Senators Clinton and Salazar, who accepted cabinet appointments.
Article I, Section 3 of the United States Constitution states that should a vacancy occur in the Senate, the executive of that state is entitled to make a temporary appointment, until the state legislature can convene to name a replacement.This was revised with the passage of the 17th Amendment and the direct election of senators to exclude state legislatures from the process.Now, if a sitting senator dies while in office or chooses to resign for any reason, the governor of that state is empowered to appoint a senator.The exception to this rule is that state legislatures may pass a law prohibiting the governor from making an appointment and only allowing the vacancy to be filled by a special election; this is currently the case in Oregon, Rhode Island, and Wisconsin.Other restrictions can be placed on the governor's appointment power.For instance, in Arizona the governor must appoint an individual of the same party as the previous senator, while in Hawaii, Utah, and Wyoming the governor is limited to names on a list provided by the previous senator's party (Neale 2013).Yet, in other states, the governor is unrestricted in his ability to appoint whomever he would like, including himself.There have been nine instances of a governor appointing himself to fill a vacant Senate seat.In all but one of these instances the governor lost the first time he had to run for election to the seat 2 (Morris and Marz 1981).

The Incumbency Advantage
When seeking reelection, countless studies and data show that incumbents, the individual currently holding the office, are overwhelmingly successful in their quests to maintain the office.The presence of an incumbency advantage in elections to the House of Representatives is well documented.Jacobson and Carson (2016) find that House incumbents from 1946 to 2014 have an average reelection rate of 92 percent.The incumbency advantage in the Senate is notably lower, with a reelection rate of 80 percent over the same time period.However, this is still a significant advantage, and given the choice between running as a challenger, incumbent, or in an open seat, any knowledgeable politician would choose to run as an incumbent.
While a complete review of the incumbency advantage literature is beyond the scope of this paper, what follows is a brief highlight of some of the proposed causes.The cause of the relative safety of members of Congress seeking reelection has been a fruitful field of research in political science.Some propose that the office itself is set up in such a way that it benefits incumbents' reelection needs.Mayhew (1974a) argues that "if a group of planners sat down and tried to design a pair of American national assemblies with the goal of serving members' electoral needs year in and year out, they would be hard pressed to improve on what exists" (p.81).While Congress as a whole may be reviled, voters often express approval of their individual member.Mayhew (1974b) suggests that the ability of incumbents to advertise, credit-claim, and take advantageous positions may be responsible for the incumbency advantage.
Incumbents have significantly higher name recognition than other candidates for office.When asked to identify the name of an incumbent, constituents are often able to name the individual but struggle to recognize the name of a challenger during an election season (Mann and Wolfinger 1980).The ability to perform constituent service and reach out to voters through free mailings provided by the franking privilege helps a member to increase their name recognition (Cover and Brumberg 1982;Cox andMorgenstern 1993, 1995).Another proposed cause of the incumbency advantage is the assistance that elected officials can provide their constituents in navigating the red tape of the federal bureaucracy (Fiorina 1977).In a similar vein, incumbents are also uniquely suited to secure federal funding for grants and pork-barrel projects at home in their districts (Bickers and Stein 1994).Finally, Prior (2006) argues that the incumbency advantage can be traced to the growth in television.As incumbents are able to receive more positive news coverage and reach constituents through television, they have been able to aid in their quests for reelection.
It is important to highlight that more recent work has noted a decline in the incumbency advantage.Jacobson (2015) uses both survey and aggregate data to detail a decline in the value of incumbency for members of the House of Representatives seeking reelection.The incumbency advantage increased from the 1950s into the 1980s; however, there is compelling data to suggest its decline since that time.Jacobson argues that the timeline of this decline suggests that the weakening of party loyalty was the primary cause of the incumbency advantage.For our purposes here, it is worth noting that Jacobson notes a similar, but not as pronounced, decline in the incumbency advantage for senators over the same period.

The Incumbency Advantage in the Senate
While a substantial majority of the research about the incumbency advantage focuses on elections to the House of Representatives, there have been a number of scholars who have examined the phenomenon in the Senate.Much of the Senate research reframes the question slightly asking not about the advantage enjoyed by incumbents, but why Senate elections are more competitive than House elections.Abramowitz and Segal (1993) examine voters in Senate elections, the primary process, election outcomes, and campaign spending.Repeatedly, their findings point to the importance of the challenger in Senate elections.Senate seats are frequently considered more prestigious and they are not up for election as often, so they are more likely to attract skilled politicians or well-known "celebrities".Cannon (1990) identifies these "actors, athletes and astronauts" and claims that they have the ability to raise large sums of money necessary to mount a credible Senate campaign.Similarly, Krasno (1994) compares House and Senate races and concludes that senators are more likely to lose a reelection bid because of the opponents they face.
In each election to the House of Representatives, candidates have roughly the same number of constituents to appeal to.However, in Senate elections the constituency can vary wildly from less than one million to thirty-eight million in the most populous state of California.Contests in small states are found to be less competitive than large states whether an incumbent is running or it is an open seat race (Lee and Oppenheimer 1999).Similarly, Adams and Squire (1997) find that senators from larger states are more likely to attract a quality-challenger, because there is a greater pool of eligible quality challengers in those states.Highton (2000) offers one of the most comprehensive examinations of the incumbency advantage in the Senate to date.He tests three critical variables: state partisanship, incumbency, and national tides.Importantly, Highton does not assume that these variables have a consistent effect over time.Examining senate elections since the passage of the 17th Amendment, he finds that the effect of incumbency has not been constant over time.He instead finds that the incumbency advantage has grown steadily since the end of the Second World War, and finds that without the advantages of incumbency, 35 percent of incumbents would have lost (p.506).
The literature on the incumbency advantage in the House and Senate points to any number of causes: name recognition, constituency service, challenger quality, the advent of television, and the franking privilege.Still, other scholars talk about a direct and indirect effect of incumbency (Cox and Katz 1996).Some of the power of incumbency lies not just in the work done while in office but from the ability to scare off potential challengers just by your presence.
Studies of the incumbency advantage tend to examine the electoral fates of the individual currently holding the office in his or her quest for reelection.Appointed senators fall into a unique category as they hold the office without ever having won an election to that position.Morris and Marz (1981) examine the fates of appointed senators from 1945 to 1979.They find that appointed senators are only half as likely as their elected counterparts to win election.In fact, a significant percentage of those appointed senators who seek to retain their seats lose in the primary and never make it to the general election.They also find that when the governor tries to appoint a senator of a different party than the previous senator, that new individual fares much worse in the election.Finally, they study the committee assignments given to appointed senators.These newly appointed senators do not assume the assignments of their predecessors; instead they are frequently given the least prestigious assignments in the chamber.They conclude that "it is surely a good thing to be a United States senator, but it is just as surely better, or at least more secure, to become one through election rather than appointment" (p.80).

Theory
While these appointed senators may, at first glance, look like incumbent candidates, the limited time they are able to serve before facing the electorate limits the advantages they are able to accrue.Appointed senators certainly may enjoy some increased name recognition.However, their ability to capitalize on many of the other advantages is curtailed by their limited time in office.Mayhew (1974a) highlights three activities that members of Congress engage in to improve their chances of being reelected: advertising, credit claiming, and position taking.In particular, newly appointed senators do not have much time to accomplish legislative goals for which they can claim credit.In addition to legislative success, credit claiming also includes constituency service.Appointed senators have limited opportunities to assist constituents.It could also be that constituents simply turn to the more senior senator from their state when seeking assistance and thus the appointed senators is not afforded the same opportunity to credit-claim.
Another factor that may affect appointed senators differently is how potential candidates view the race.Cox and Katz (1996) identify a scare-off effect that helps incumbents by discouraging quality challengers from entering the race in the first place.I propose that appointed senators likely do not benefit from the same scare off effect.In fact, they may be more likely to attract quality candidates and in more recent elections see more outside money poured into their race.Potential candidates may view the race more like an open seat than an incumbent-held seat.This effect could be present both in the primary and general as members of the appointed senators own party likely do not give them the same deference they would a six-year incumbent.
A brief glance at the list of appointed senators tells us that not all aim to seek reelection.Notably, the wives of several deceased senators served as placeholders until an election could be held.This means that not all appointed senators fall under the guiding theory of congressional elections research, that members of Congress are "single-minded seekers" of reelection (Mayhew 1974a, p. 17).However, there are also examples of appointed senators who went on to serve lengthy terms in office.Harry Flood Byrd (D-VA) was appointed to his Senate seat, won his initial reelection battle, and went on to serve for more than thirty years.In the 115th Congress, there are eleven senators serving who first came to the chamber as appointed senators to fill a vacancy.
Those appointed senators who do survive the primary election and seek to maintain their seats perhaps should be treated as quasi-incumbents.They have more advantages than a candidate challenging a sitting incumbent or a candidate running for an open seat.However, they also must juggle a compressed time frame prior to the election.Not only must they get a senate office up and running, but if they wish to be reelected they must quickly create a campaign operation.Sitting senators have five years prior to the election year to fundraise and plan for their campaign.
Nevertheless, studying the decisions and electoral effects of running as an appointed senator is important to gain a more complete understanding of Senate elections.It is also notable as an examination that could possibly shed some light on the causes of the incumbency advantage.Comparing appointed senators to first-term senators does not provide a perfect natural experiment; however, it does allow an interesting study of how those who have never won the office fare.Are they able to marshal enough advantages of the office to win at the same rates or does the limited time they are able to serve prevent them from truly accruing a significant incumbency advantage?

Data
To examine the differences between electoral outcomes for appointed and first-term senators, I use data from 1920 to 2012.The direct election of senators began with the passage of the 17th Amendment and the 1914 election.I exclude the first three years in which all senators were directly elected so that those seeking reelection for the first time will have previously gained their seats through direct election rather than being selected via state legislature.Included in the dataset are 142 appointed senators and 483 Senators who have completed their first six-year term in office and are seeking reelection for the first time.Those appointed senators who, because of state law, faced a special election held outside of the regularly scheduled November election date were excluded from the dataset.To compare the electoral fates of senators who had served multiple terms in office to newly appointed individuals who have never been elected to the Senate highlights too stark a difference.It would be unfair to compare the electoral fates of newly appointed senators to a senator who had served his state for thirty-plus years.A complete list of appointed senators is available on the Senate website (United States Senate n.d.).Not all are included in the dataset as some served under unique circumstances, such as Rebecca Latimer Felton (D-GA), who was appointed and served in the Senate for only 24 h.As the first woman to serve in the Senate, her appointment was largely symbolic.There are other unique circumstances that led to exclusion from the dataset such as individuals appointed to fill a vacancy who died before the election was held (William Crow in 1921).Data on the election of all senators comes from Michael Dubin's United States Congressional Elections, 1788-1997: The Official Results.Additional information came from the Congressional Biographical Directory online (http://bioguide.congress.gov/biosearch/biosearch.asp).
When studying the decision to run for election or electoral outcomes, I rely on a variety of control measures.The control variables included capture measures of the state and measures specific to the senate seat.Here, I first describe the state variables.Population has proven to be an important measure in Senate races as campaigning in California is dramatically different than running a campaign in Vermont.Squire (1991) finds that the state population impacts the size of the pool of eligible candidates, which in turn impacts the emergence of a quality challenger.I use state population measures from the U.S. Census Bureau.I also control for southern states because of the large timeframe used in the study; for a large portion of the 20th century, elections in southern states were uniformly won by Democratic candidates.In addition, I control for the solid south by using a dummy variable for the 11 states of the confederacy. 3For large portions of the period covered in this analysis, the Democratic Party dominated politics in the southern states almost completely.This could have important effects on both the decision to seek the office or the electoral success of a candidate.Another critical control variable is the presidential vote in that state.I use the presidential vote of the candidate of the same party as the incumbent senator in the most recent presidential election.This variable serves as a measure of the partisanship in the state.Information on presidential election results by states comes from Dave Leip's Atlas of U.S. Presidential Elections (http://uselectionatlas.org/).
I turn now to the control variables specific to the senator.I control for the age of the appointed senator and the length of service in the Senate.One might expect that older appointees are more likely to be chosen by governors to serve as place holders.The appointed senators in the dataset varied in age from 34 to 79 with a mean age of 54.In addition, appointees must run for the recently vacated seat at the next election, so at most they can serve up to two years before they must be elected to the office.The length of service variable is measured by the number of months the appointed senator served prior to Election Day.There is certainly a different calculus for those appointed senators who serve a full two years as opposed to those who serve just a few months prior to the election.Here the average length of service for senators in the dataset is twelve months with variation from one month prior to election to twenty-four.
Finally, a control is included for those newly appointed senators who are from a different party than the senator they are replacing.In many states, governors are free to appoint whomever they wish to serve in the seat so long as they meet the basic age and residency requirements as laid out in the Constitution.As a result, the appointment of a new senator by a governor of the opposing party often flips the seat to the opposing party.While the party at large views the pick-up of a new Senate seat as an advantage it may not be an advantage for the appointed senator.The change in party control of the seat may draw additional scrutiny to the appointed individual in his or her quest for reelection.In the dataset, only 22 percent of the appointments altered the party control of the Senate seat.The party change variable is coded one if the newly appointed senator is of a different party then the senator who vacated the seat.
The final control variable is for is whether the appointed senator has ever previously held an elected office.Figure 1 provides descriptive statistics about the previous careers of these 146 appointed Senators.Of these 146 appointed senators, only fifty-four percent had ever previously held elected office.The plurality of those individuals were appointed from state legislatures.It is worth noting that of the seventeen governors who were appointed to the Senate, seven appointed themselves to the position.However, the self-appointment of governors happened from 1940 to 1966.No governor since that time has taken the political risk of appointing himself to the seat.Those appointed senators with no previous elected experience primarily had either a legal or business background.Within this dataset there are six women who were appointed to fill the vacancies created by the death of their husbands.A few came from appointed political offices or had served in the government in some other non-elected capacity.I turn now to the control variables specific to the senator.I control for the age of the appointed senator and the length of service in the Senate.One might expect that older appointees are more likely to be chosen by governors to serve as place holders.The appointed senators in the dataset varied in age from 34 to 79 with a mean age of 54.In addition, appointees must run for the recently vacated seat at the next election, so at most they can serve up to two years before they must be elected to the office.The length of service variable is measured by the number of months the appointed senator served prior to Election Day.There is certainly a different calculus for those appointed senators who serve a full two years as opposed to those who serve just a few months prior to the election.Here the average length of service for senators in the dataset is twelve months with variation from one month prior to election to twenty-four.
Finally, a control is included for those newly appointed senators who are from a different party than the senator they are replacing.In many states, governors are free to appoint whomever they wish to serve in the seat so long as they meet the basic age and residency requirements as laid out in the Constitution.As a result, the appointment of a new senator by a governor of the opposing party often flips the seat to the opposing party.While the party at large views the pick-up of a new Senate seat as an advantage it may not be an advantage for the appointed senator.The change in party control of the seat may draw additional scrutiny to the appointed individual in his or her quest for reelection.In the dataset, only 22 percent of the appointments altered the party control of the Senate seat.The party change variable is coded one if the newly appointed senator is of a different party then the senator who vacated the seat.
The final control variable is for is whether the appointed senator has ever previously held an elected office.Figure 1 provides descriptive statistics about the previous careers of these 146 appointed Senators.Of these 146 appointed senators, only fifty-four percent had ever previously held elected office.The plurality of those individuals were appointed from state legislatures.It is worth noting that of the seventeen governors who were appointed to the Senate, seven appointed themselves to the position.However, the self-appointment of governors happened from 1940 to 1966.No governor since that time has taken the political risk of appointing himself to the seat.Those appointed senators with no previous elected experience primarily had either a legal or business background.Within this dataset there are six women who were appointed to fill the vacancies created by the death of their husbands.A few came from appointed political offices or had served in the government in some other non-elected capacity.

Results
First, I examine the decision to seek reelection.Rohde (1979) posits that all members of the House of Representatives have progressive ambition, meaning that if it were costless, they would choose to seek higher office.For senators, understanding ambition is more complicated.There is no natural next step for a senator.Some seek the presidency, some may choose to run for governor or hope to be appointed to a cabinet-level position, and still others may simply want to remain in the Senate for as long as possible.This desire to simply maintain their seat is what Rohde would classify as static ambition.Appointed senators may not have static ambition, they may simply want to hold the seat for a year or so and then return to their previous endeavors.Morris and Marz (1981) point out that some appointed senators are merely "seat warmers" who likely promise the governor they will not seek reelection.
Table 1 outlines the reelection decisions of the appointed and first-term senators.Looking first at the appointed senators, it is clear that many serve as seat warmers or simply decide they would prefer not to put themselves through the grueling campaign process.Thirty-three percent of the appointed senators in the sample simply chose to retire and not seek election.Another 13 percent sought election, but lost the nomination for the seat they were currently holding.This leaves only 54 percent of appointed senators who sought to keep the seat for which they had been appointed in the general election.The first-term senators clearly arrive with more uniform goals and most seek reelection (88.6 percent).A mere 3.5 percent of first-term senators lost their nomination contest.

Number of Cases in Parentheses
In order to first examine what conditions make an appointed senator more likely to seek reelection, I ran a logistic regression controlling for a variety of factors.The dependent variable is coded one for all appointed Senators who sought reelection including those who lost in the primary, because it was their desire to maintain their seat if possible.It is coded zero for all those who chose to retire and never campaigned for the office.The first independent variable, the vote share of the presidential candidate of the appointed senator's party, was included as a measure of the partisanship of the state.A control was also included for southern states because of the one-sided nature of southern politics throughout much of the time studied.In addition, a control for female appointed senators was included as many of them were appointed following the death of their husbands purely as place holders.Some of these women went on to seek the office and others quickly retired to return to private life.I also include a control for party to determine if there are any party-specific effects at play.Finally, state population is included because senators face wildly differently sized constituencies.An appointed senator from a larger state may fear the larger pool of eligible candidates who could potentially challenge them for the seat.Controls were also included for the age of the appointed senator and for the number of months that individual served in the Senate prior to election day.A variable was included to capture whether or not the newly appointed senator caused a party change in the Senate seat.Finally, as a measure of candidate quality of the appointed senators a control was added for those who had previously held an elected office.
The results of the logistic regression are presented in Table 2. Interestingly, Democrats were less likely to seek reelection to an appointed Senate seat than their Republican counterparts all else being equal.Age is also a statistically significant predictor of the decision to seek election to the Senate.Older appointed senators are less likely to seek reelection in their own right.This is unsurprising as it may be a way for governors to ensure that appointed senators serve as place holders.There are several instances of former senators being reappointed to serve in the Senate for this brief period.The length of service in the Senate is positive and significant.Those who are appointed earlier in a Congress and are able to serve longer are more likely to seek the office.Again, this likely speaks to the strategic nature of the appointments made by governors.They are aware that given a two-year lead time an individual is more likely to be able to adequately serve in the Senate and gear up for a campaign to keep the seat.Whereas those who are appointed for shorter amounts of time do not have the ability to adequately fundraise and mount a credible campaign for the office.
Those appointed senators who had previously held an elected office were more likely to seek to maintain the seat by campaigning for reelection.These candidates have already demonstrated an ability to win an election with most of them serving either in a state legislature, the House of Representatives or as a governor.Previously holding an elected office has repeatedly proven itself a key predictor of electoral success for candidates (Jacobson and Kernell 1981).
Surprisingly, variables to capture the partisan nature of the state or whether or not a party change occurred are not statistically significant. 4Each individual appointed to the Senate comes to the office because of unique and often unforeseen circumstance.Certainly the decision whether or not to seek election to the office after having been appointed can be idiosyncratic and each case is unique.However, these findings could help predict which appointed senators in the future are most likely to seek to stay in the Senate.Next, I turn to a consideration of the electoral fates of appointed senators and first-term incumbents.Table 3 presents the percentage of candidates who won and lost for the two groups.A larger percentage of the first-term senators won reelection as compared to the appointed senators.The reelection rate for first-term senators in the sample (79.3 percent) is extraordinarily close to the reelection rate typically given for senators as a whole in the post-World War II era.The win-loss difference between the two groups of senators was found to be significant at the 0.10 level.Interestingly, the mean vote share for each of these groups is remarkably similar.The mean vote share for winning appointed senators is 64 percent while the mean for first-term senators is 63 percent.For losing candidates the vote shares are nearly identical with both groups losing with 46 percent of the vote.Difference is significant at the p < 0.10 level.The number of cases is in parentheses.
Initially, it seems that appointed and first-term senators face similar electoral outcomes, although first-term senators do have a slightly higher reelection rate.Yet this does not account for other factors that we know influence election outcomes.Models One and Two of Table 4 presents an Ordinary Least Squares regression where the dependent variable is the incumbent candidates vote share.Model Three presents a logistic regression where the dependent variable is coded one if the senator won reelection and zero if they lost.I have excluded 24 observations where the incumbent or appointed senator ran unopposed in the general election.Obviously, the key variable of interest is whether or not the senator was appointed.To account for the partisanship of the state the vote share of the presidential candidate of the same party as the incumbent in the most recent presidential election is included.Dummy variables are included for Democrats and senators from southern states.Finally, a variable is included to measure the quality of the opponent the incumbent faced.Candidate quality has proven to be a key variable in studies of Senate elections (Krasno 1994).Here, I use the Jacobson and Kernell (1981) measure most frequently used in the House of Representatives coding challengers as a quality opponent if they have previously held any elective office and zero if they have not.Unfortunately, candidate quality information can be difficult to obtain, particularly for the earlier Senate races, causing some cases to be excluded from the analysis.Model One presents the results with only those individuals whose candidate quality information could be located.For Model Two, I have recoded those with missing challenger quality scores as zero.This assumption is frequently made that if candidate quality information cannot be located then the individual likely has never held an elected office (Jacobson 1989).Certainly, they did not serve in the House of Representatives because they could not be located in the Congressional Biographical Directory or various other sources.
Table 4 shows that in both models the effect of being an appointed senator, as opposed to having won a seat six years prior, is statistically significant and negative.Controlling for other factors, being an appointed senator can lower your vote share from 2.26 to 3.65 percentage points.Appointed senators only hold the office for a maximum of two years before they must face the voters in a special election to secure the seat.It seems those two years in office are not sufficient to secure the same advantage as being elected to the office.The vote share of the presidential candidate of the incumbent senator's party is a positive and significant predictor of the incumbent's vote share.This is expected since the presidential vote can serve as an indicator of the state's partisanship, and we would anticipate that senators would have a higher vote share in states that tend to vote for candidates of the incumbent's party in other offices as well.In controlling for party, Democrats had a significantly larger vote share.This could be due to the time period studied, which included a period of substantial one party control.Similarly, senators from states located in the south had higher percentages of the two party vote.Again, this is likely due to the one party dominance of politics in the south.Finally, as expected the presence of a quality challenger has a significant negative effect on the incumbent's vote share in both models.Quality challengers who have previously held elected office know how to run and manage a campaign and may already have some level of name recognition among the electorate.
The logit results paint a similar picture for appointed senators.All of the variables remain statistically significant and in the same direction with the exception of the dummy variable for southern states.Appointed senators were thirteen percent are less likely to win than first-term senators holding all other variables constant at their median or modal value.These findings suggest that while there is some electoral advantage to holding the office, serving as an appointed senator is in no way a guarantee that an individual will be successful when they seek reelection to the Senate.The percentage of appointed senators winning reelection was close but less than that of first-term senators and their mean vote share was nearly identical.However, once other factors were controlled for, being an appointed senator had a negative impact on an incumbent's share of the two party vote and the likelihood of winning the election.

Discussion and Conclusions
After Barack Obama won the presidency in 2008, he resigned his Senate seat, leaving it to the Governor of Illinois to fill his seat.Then-Governor Rod Blagojevich, who is now known for attempting to sell off the Senate seat, was recorded by federal investigators using expletives to describe the vacant seat as "golden".He viewed the ability to hold the Senate seat as such a rare gift to someone's political career that he planned to get something for himself in the appointment process.While Blagojevich's story is largely about corruption, it highlights the perceived value of being appointed to the U.S. Senate.However, the reality of those appointments can be different.
Of course, Senate seats are highly valued.Typically, they are only up for election every six years and are one of the more prestigious statewide offices a politician seeks.The ability to serve in the Senate, gain name recognition, and serve constituents as a sitting senator would seem to be extremely appealing to ambitious politicians.Yet only sixty-seven percent of the appointed senators chose to run in an election in an attempt to keep that seat.Of course, some governors appoint individuals knowing that they do not intend to continue on in the Senate.There are several instances since the passage of the 17th Amendment when a governor appoints the wife of a deceased senator to finish out the term.
Predicting who will seek re-election and who will simply serve out the reminder of the term until an election can be held is challenging.However, this analysis suggests that older appointed senators are less likely to campaign to keep the seat.When the appointed senator takes over the office is also an important predictive variable.Those who are able to take the appointed seat with a greater amount of time prior to the election are more likely to run for reelection, whereas those who are appointed with limited time until election day are more likely to serve as place holders and then step aside.In addition, those appointed senators who had previously held an elected office were more likely to seek to maintain the seat.Many of these individuals have already built a career in public office and they know how to run and manage a winning campaign.
Turning to the electoral success of appointed senators, an initial examination of descriptive statistics suggested that they fared similarly, if not slightly worse to other first-term senators.However, once other factors were controlled for it seems that running as an appointed senator in fact has a negative impact on the incumbent's vote share all else equal.Admittedly, an appointed senator has less time in office to spend engaged in the "advertising, credit claiming, and position taking" activities first outlined by Mayhew (1974a).However, the electoral advantage enjoyed by senators who had previously won the office in a regularly scheduled election suggests that perhaps it is not the trappings of the office, but skill in electioneering that contributes more to the incumbency advantage.
This idea that experience running a winning campaign is what contributes to the incumbency advantage is backed up by the importance of quality challengers.We know that challengers who have previously won elected office present much stiffer competition for incumbents.Time and again it has proven to be a critical measure in predicting electoral outcomes.Perhaps the driving force of the incumbency advantage truly is the experience of running and winning a campaign.Not all candidates are created equal; some are more natural at giving speeches, kissing babies, and asking for support.Candidates who have previously won have demonstrated that they can be successful.Similarly, work by Steen (2006) finds that money is a necessary but not sufficient condition for winning elections, suggesting that electioneering skill is critical.When it comes time to run for office again, they already know how to build a campaign staff, raise money, and compete.Leveraging the unique circumstances faced by appointed senators provided us with a small insight into the incumbency advantage.Again, there are likely many contributing factors to the incumbency advantage in the Senate, but time and again research suggests that previously having won an election is a critical predictor of future success.
The only way to truly isolate whether the advantage enjoyed by incumbents stems from activities while in office or electioneering skill would be if appointed senators served full six-year terms.Because appointed senators must face their constituents in an election within two years' time, they will always be at a disadvantage as compared to other first-term senators.
Running as an appointed senator may in fact be both a blessing and a curse for those individuals.Serving as the sitting senator confers upon the appointee increased name recognition and the status of serving as an incumbent, yet these individuals do not perform as well electorally as other first-term senators.Perhaps simply serving in the office is not enough.Contrary to the colorful sentiments expressed by Governor Blagojevich, the appointment to a seat provides no guarantee that its new inhabitant will be able to retain the position.In fact, they may be better off allowing a seat warmer to hold the office and then run in an open seat race.

Table 1 .
The Decision to Seek Reelection for Appointed and First-Term Senators.

Table 2 .
Logit Predicting Whether or Not an Appointed Senator Runs for the Office.

(Robust Standard Error Clustered by Year)
4Running the model clustering by states and with election year fixed effects yielded similar results.

Table 3 .
Percentage of Candidates Winning Reelection.

Table 4 .
Election Results Models.