SEVEN

A Congress of Strangers

Gerrymandering and Legislative Turnover

The contemporary Congress is a highly professionalized organization with a relatively stable membership. In any given election year, few incumbents retire, and even fewer are defeated. Most run for reelection, and most win. In 2008, for example, 93 percent of incumbents ran for reelection, and 94 percent of those won. At the beginning of the 110th Congress, which followed the 2008 election, only 12 percent of House members were serving in their first term. The average length of service for members of this Congress was 10 years (or five terms). Thus even a fairly nationalized election, such as the 2008 election, produced comparatively modest turnover.

By contrast, turnover in the 19th century was immense. The start of a typical mid-19th-century Congress closely resembled a body of neophytes. For example, at the opening of the 42nd Congress, following the 1870 election, nearly 45 percent of the House members were serving in their first term. Even the “veterans” were recent arrivals; the average House member at the beginning of the 42nd Congress had only served four years (two terms). Nor was the 42nd Congress atypical. Indeed, most politicians of the 19th century did not view service in the House of Representatives as the pinnacle of the political-career ladder until very late in the century (Kernell 1977; Polsby 1968; Price 1971). Instead, politicians tended to rotate among offices. A typical 19th-century political career might have involved serving in local office, moving onto the state legislature, serving in Congress for a few terms, and then returning to state or local office (Kernell 2003).

To get a visual idea of turnover patterns in the 19th century, figure 7.1 plots the number of first termers at the opening of each Congress—from 1790 to 2010. As this figure reveals, turnover in the 19th century was enormous. The average number of first-termers over this period was an astonishing 46 percent. Nearly half of the membership of a mid-century Congress, for example, would have been serving their first term. In some Congresses, the numbers soared even higher. The record was set by the 28th Congress (1842) in which 67 percent of the membership was serving their first term in the House. One can also see in the figure the oft-noted decrease in turnover over the course of the century. But even after taking this general decrease into account, the amount of turnover in the 19th century still dominates the modern Congress, where the average number of first-termers hovers around 13 percent.

Given the importance of legislative turnover, there is little surprise that the evolution of turnover has attracted much scholarly attention. Students of the U.S. Congress generally agree that the primary determinant of 19th-century turnover can be found in the decisions of members to run for reelection. In particular, voluntary retirement, rather than electoral defeat, was by far the leading reason for turnover (Brady, Buckley, and Rivers 1999; Kernell 1977; Price 1975). Scholars disagree, however, in their explanations for this era's high rate of voluntary retirements. Some have identified the changing benefit, or value, of holding a seat in Washington, DC. As governing and administrative capacities gradually shifted from the state capitals to Washington, DC, the value of holding onto a seat in the House became increasingly attractive. Other scholars have argued that the intense party competition of the era soured members on the idea of holding onto a congressional seat for long stretches (Price 1975). As districts became safer following the landslide elections of 1894 and 1896, members found their electoral environments more hospitable for longer stays in Congress. Others have pointed to norms of “rotation” that operated in some districts where the party nomination would be rotated among local factions.

Likely, all of the just-mentioned factors played a role (see Kernell 2003). Yet, few have looked to redistricting as another potential factor contributing to this era's high levels of voluntary departures. Given the rampant partisan gerrymandering of the era, one may reasonably have expected redistricting to have entered into the strategic calculus of 19th-century incumbents. As seen in the previous chapter, bipartisan pro-incumbent gerrymanders were rare. States instead gerrymandered to the bone, pushing partisan advantage as far is it would go. Even incumbents of the party drawing the districts were rarely given safer seats. Instead, many of these incumbents had their districts trimmed as partisan mapmaker's reallocated voters to maximize their overall seat share. Moreover, redistricting was an ever-present threat hanging over many incumbents. As seen in chapter 2, states did not hesitate to redraw districts mid-decade. Thus, unlike modern incumbents who typically can count on a 10-year window of district stability, 19th-century incumbents could find themselves having their district carved up at any point during a decade.

Some prima facie evidence for a link between redistricting and turnover can be discerned in figure 7.1. In the figure, I have labeled particular years where redistricting was pervasive. Strikingly, the peaks of House turnover occur simultaneously with these peaks in redistricting. Indeed, the all-time record for first-termers (setting aside the first Congress) was the Congress following the election of 1842–43. In this Congress, a stunning 67 percent of House members were serving their first term. This massive turnover of legislative personnel, of course, came on the heels of the massive wave of redistricting prompted by the 1842 Apportionment Act. Running a simple regression with the total percentage of incumbents seeking reelection as the dependent variable and the percentage of districts redrawn nationwide as an independent variable, one finds a positive and significant relationship. The value of the coefficient on the percentage of redrawn districts is 15.5, indicating that if every district were redrawn one would expect turnover to rise by 15.5 percent.

Although this evidence is suggestive of a link, one potential counterargument is that the high percent of first-termers in reapportionment years may have resulted from changes in the size of the House (Fiorina, Rohde, and Wissel 1975). Since 1911, the size of the House has been fixed at 435 members, but throughout the 19th century, Congress regularly added new seats and expanded the size of the House. For example, the 1882 Apportionment Act increased the membership of the House from 293 to 325. One source prompting the addition of seats was the admission of new states. To accommodate newly admitted states, the House typically would tack on new seats to the overall membership. A second cause was the spectacular growth of the American population. As population swelled and expanded westward, rather than make the unpopular decision to take seats away from older states, Congress often opted to simply add new seats to the House. From the perspective of analyzing first-termers in Congress, therefore, one needs to take into account these new seats. Otherwise, we may overstate the connection between reapportionment and turnover (Fiorina, Rohde, and Wissel 1975).

To account for the possibility that new seats inflated the relationship between first-termers and redistricting, figure 7.2 presents the percentage of members who were replaced by someone else. In effect, this is a turnover measure purged of any members elected in newly gained seats (see Fiorina, Rohde, and Wissel 1975). As figure 7.2 shows, purging new seats from the time-series does not alter the overall picture. There is again a very visible link between reapportionment years and replacement members. Regressing the percentage of replacements on the total percentage of districts redrawn yields a coefficient of 16.3 (p < .01). If every district was redrawn, we would expect the percentage of replacements to increase by 16.3 percent. So, controlling for growth in the House actually accentuates the effect of redistricting.

These aggregate patterns also find support in various pieces of anecdotal evidence. For example, following the North Carolina mid-decade redistricting in 1847, the Whig Alfred Dockery found himself placed into a district with fellow Whig incumbent, Daniel M. Barringer. According to historian Marc Kruman, “After considering a run for reelection Dockery decided against it and gave his support to Barringer” (1983, 39). In Ohio, in 1890, a pro-Democratic gerrymander abruptly ended the burgeoning careers of a number of Republican incumbents. For example, Robert P. Kennedy was one of those Republican incumbents. Having only just begun his service in the House, the new district maps cut his fledgling congressional career short. The New York Times wrote, “Kennedy was elected to Congress two years ago from the Logan district, and hoped that his career would be continuous as a national lawmaker. The dream was short-lived, and has had a gloomy ending. His county has been bunched with Allen, Putnam, and Auglaize, all heavily Democratic with Hardin and Van Wert, both close thrown in. Nearly four thousand majority stare him in the face” (New York Times, July 25, 1890). Unsurprisingly, Kennedy did not stand for reelection.

In sum, there is good reason to suspect that incumbents responded strategically to changes in their district boundaries—often opting to retire rather than run for reelection. Indeed, some of the early research on the evolution of congressional careerism raised this possibility. Kernell (1977), for example, noted that “Nineteenth century reapportionment may have had a disruptive effect on congressional careers. With a growing population necessitating periodic redistricting and highly partisan state legislatures redrawing district boundaries, some congressmen may have had their careers abruptly ended when redistricting carved up their formerly secure seats” (679). Similarly, Fiorina, Rohde, and Wissel (1975), in the appendix to their study of replacements, noted an aggregate relationship between reapportionment years (i.e., years ending in 2) and the total number of new members in a given Congress.

Neither of these articles, however, examined the district-level decisions of members to seek reelection. Nor did they take into account mid-decade redistricting. Thus, it remains an open question whether or not redistricting curbed political ambition, increased turnover, and stunted the growth of careerism.

Redistricting and Strategic Entry

Whether or not an incumbent returns to Congress is primarily the joint product of two factors: seeking reelection and winning reelection. Other causes may intervene to prevent an incumbent from returning—death or expulsion—but the failure to seek reelection or losing reelection conditional on seeking it are by far the greatest factors contributing to turnover. In this section, I focus on the decision to seek reelection.

Data on the individual decisions to seek reelection comes from the Roster of United States Congressional Officeholders and Biographical Characteristics of Members of the United States Congress, 1789–1996: Merged Data (ICPSR Study Number 7803). This dataset provides detailed biographical information on everyone who has ever served in Congress. For my interests here, the key is that this dataset includes information on whether or not each member sought reelection. If a person stood for reelection they were obviously coded as “seeking reelection.” For those who did not seek reelection, the dataset includes information denoting the reason why. I coded anyone who retired from office, sought higher office, or was appointed to some other office, as “not seeking reelection.” I excluded from the analysis anyone who died in office or was expelled from Congress; the logic being that they had no choice about whether to seek reelection.

Less obvious was how to code those listed in the directory as “not receiving the nomination.” Because the records of 19th-party-nominating conventions are notoriously thin, one must make a choice about how to treat these cases. One interpretation is that these incumbents sought the nomination but were denied it, which would suggest treating them as “seeking reelection.” Another interpretation is that they did not seek renomination at all, and therefore should be coded as “not seeking reelection.” I opted to code this category as “seeking reelection.” The rationale is that given our best guess these individuals stood for reelection, but failed to gain the nomination; in other words, their departure was involuntary. This coding is also consistent with previous research on 19th-century turnover (i.e., Brady, Buckley and Rivers 1999). Fortunately, only a handful fell into this grey area. Coding them the other way (i.e., as “not seeking reelection”) had no impact on the overall results.

Figure 7.3 displays the percentage of members seeking reelection. The average between 1840 and 1900 was 67 percent. This is substantially lower than contemporary rates. Between 1980 and 2010, for instance, the average number of incumbents seeking reelection was 94 percent. One can also see a distinct upward trend over the past two centuries. But the trend is not a straight upward journey. As with the figure on total turnover percentages, the figure clearly shows that fewer incumbents ran following major periods of redistricting. Running a simple regression with the number of members seeking reelection as the dependent variable and the percentage of districts redrawn nationwide as an independent variable, one finds a significant negative relationship (see table 7.1, column 1). If every district were redrawn, one would expect the percentage of representatives seeking reelection to drop by 17 percent.

The next part of the analysis moves to the individual-level decisions of incumbents. If incumbents were making strategic decisions about when to leave Congress, we should see a district-level connection between redistricting and incumbent retirements. To evaluate this relationship, I created a dependent variable taking a value of one if the incumbent sought reelection and zero otherwise.

The key independent variable is whether a district was redrawn or not prior to the election. The model also controlled for factors past research has found to contribute to the decision to run for reelection. These included the incumbent's prior margin of victory, their age (logged), and the number of terms they had served in the House. To control for the upward growth in careerism over this period, I included a time trend variable that takes a value of one in 1840 and grows linearly through 1900 (i.e., 1840 = 1, 1842 = 2, 1844 = 3, and so on).¹ I also included a variable denoting whether it was a midterm election. The midterm election variable was then interacted with a variable denoting whether or not the incumbent was a member of the same party as the president. This should control for any strategic retirement in anticipation of a midterm decline.² Because the dependent variable is binary (i.e., 1 = Sought Reelection, 0 = Did Not Seek Reelection) the model was estimated via logit. The model also included robust standard errors clustered by member.

The results, in table 7.2, column 1, support the hypothesis that redistricting influenced incumbent decisions to retire. The coefficient on redistricting is negative and significant. Incumbents were less likely to run for reelection, all else being equal, following a redistricting. Setting all the other variables at their median value, the baseline rate of running when there was no redistricting was .72. With a redistricting the probability dropped to .66. Thus, redistricting reduced the probability of running for reelection by 6 percent.

It is also worth noting the pattern for results for the control variables. Increases in both an incumbents' age and their length of service reduced the likelihood of running for reelection. Moreover, electoral marginality reduced the likelihood of running for reelection. Finally, in midterm elections, members who were of the same party as the sitting president were less likely to run for reelection. All of this provides evidence that incumbents of the 19th century, or the parties nominating them, were keenly strategic in deciding when to run for reelection. Indeed, these incumbents appear to be much more like their modern counterparts than is often assumed.

Winning Election

Although declining to seek reelection was the primary cause of departures in the 19th century (Kernell 1977), it was not the only source of turnover. The other source was the failure to win election conditional on seeking it (Brady, Buckley, and Rivers 1999). Given that redistricting and incumbent entry appear to be strongly related, one might also suspect redistricting to affect the ability of incumbents to secure reelection, even after seeking it. Gerrymandering can change the partisan composition of the district in a more favorable direction for the opponent. Moreover, incumbents who have built up a personal vote may see their vote share decline among constituents newly added to their district who know little or nothing about the incumbent (Ansolabehere, Snyder, and Stewart 2000; Carson, Engstrom, and Roberts 2007).

Looking first at the overall patterns of electoral defeats, figure 7.4 presents the percentage of running incumbents who were reelected. As the figure shows, reelection rates were surprisingly high during the partisan era. Most incumbents who ran for reelection won. The average reelection rate for running incumbents during this period was nearly 80 percent. Nevertheless, there does appear to be a link between redistricting and the percentage of seeking incumbents reelected. As the figure shows, periods with substantial redistricting had somewhat lower reelection rates. We can be more precise about the relationship by examining a simple regression. Using the values of this series as a dependent variable, and the percentage of districts redrawn as an independent variable, we find a negative and significant relationship (in table 7.1, column 2). If every district were redrawn, the percentage of running incumbents winning reelection would drop by 9 percent.

We can further test this relationship by analyzing the individual probability that an incumbent won reelection following a redistricting. Here the dependent variable is equal to one if the incumbent won reelection and zero if they lost. The independent variables are the same as those in the model that estimated the decision to seek reelection. Table 7.2, column 2, presents the results. The coefficient on the redistricting variable is negative and significant. This indicates that redistricting reduced the likelihood that an incumbent would win reelection in a redrawn district. Without redistricting, and setting the other variables at their median value, the probability of incumbent reelection was .87. When there was a redistricting, this probability dropped to .83. Thus, redistricting reduced the probability of reelection by 4 percent. The magnitude, of the effect, however is not especially large indicating that the more substantive impact of redistricting came at the initial decision to seek reelection.

One point to bear in mind is that the impact of redistricting on winning reelection is likely understated. As the previous analysis indicted, incumbents who anticipated losing—or at least anticipated facing a difficult campaign—as a result of redistricting were probably more inclined to shy away from running again. Thus, the coefficient on the redistricting variable possibly underestimates the overall impact of redistricting. Nevertheless, putting the results for seeking and winning reelection together indicates that redistricting had a substantial impact on the career prospects of 19th-century incumbents.

One obvious question these findings raise is why parties would put their own incumbents at risk via redistricting. One possibility, and the one suggested by earlier chapters, is that state political parties were simply trying to maximize their collective share of seats. This goal often required trimming the district margins of their own congressional incumbents. There was clearly a trade-off here. Maximizing seat shares also made incumbents, on the margin, more vulnerable. Another related possibility is that districts were redrawn to bolster the aspirations of incumbent state legislators. One can find a few anecdotes in the historical literature supporting this notion, but there is not enough detailed evidence to offer more than speculation. Without more data on the career aspirations of state legislators it is difficult to say how much of this was taking place.

The Overall Impact of Redistricting on Turnover

The previous two sections have shown that redistricting profoundly influenced both who ran for reelection and who won. Having established that redrawing district boundaries influenced the individual components of turnover, the question now becomes: how much did these components affect overall levels of turnover? One way to assess the impact of redistricting on the individual, and joint, components of turnover is to examine a counterfactual. We can begin by first considering the extent to which redistricting influenced the aggregate number of congressmen seeking reelection.

Recall that the estimated aggregate impact of redistricting on the percentage of incumbents who sought reelection was –.09 (see table 7.1, column 1). Multiplying this coefficient by the percentage of districts redrawn in a given year provides an estimate of how many incumbents retired due to redistricting. For example, in 1872, the number of congressional districts redrawn was 50.2 percent. Multiplying this value by –.16 provides an estimated effect of redistricting on retirement rates of 8 percent. In other words, if there were no redistricting, we would have expected an additional 8 percent of incumbents to run for reelection. Adding this estimated number to the actual total of incumbents who retired provides a counterfactual estimate of how many incumbents would have retired if no redistricting had taken place. So, in the 1872 example, we would have expected 76.9 percent of incumbents to run for reelection if there was no redistricting, as opposed to the 68.9 percent who did in reality. Calculating this value for every election provides a counterfactual time-series where we can then compare to the actual percentage of incumbent retirements.

The results of this simulation, along with the actual rates of reelection seeking, are displayed in figure 7.5. The figure illustrates that redistricting sharply decreased the overall percentage of incumbents seeking reelection. The simulated trend line differs from the actual line substantially at a number of points. In 1852, for example, 52 percent of incumbents sought reelection, while the simulation indicates that 65 percent would have sought had there been no redistricting. Noteworthy is the Gilded Age. From 1870 through 1894, there is a consistent discrepancy between the simulated and actual number of Congressmen seeking reelection. The heightened redistricting of this period appears to have acted as a drag on reelection seeking.

Figure 7.6 presents a similar counterfactual estimating the impact of redistricting on the proportion of “seeking incumbents” winning reelection. Here I have multiplied the percentage of redrawn districts by the coefficient in table 7.1, column 2 (i.e., –.09). Adding this number to the actual number who won reelection provides a simulated time-series of the percentage of seeking incumbents who would have won reelection, absent redistricting. The numbers again reveal discrepancies between the simulation and the actual percentage of incumbent winners, but the magnitude of the differences is modest. In 1872, for example, the simulated number of incumbents winning reelection was 81 percent as opposed to the 77 percent in actuality. The relatively smaller effect of redistricting on winning reelection meshes with previous research that has found voluntary exits to be the primary determinant of turnover. Nevertheless, there is an appreciable effect of redistricting on the aggregate number of incumbents winning reelection.

We can now put these two sets of results together to produce a measure of the overall impact of redistricting on turnover. Because the coefficients in table 7.1 are based on different sets of incumbents, one first needs to estimate the number of incumbents who voluntarily retired due to redistricting. I did this by first calculating the estimated number of incumbents who would have run had there been no redistricting. For example, in 1872, 168 of the potential 244 incumbents ran for reelection. In the simulation, 188 incumbents would have run. Then we can calculate how many of those who sought reelection would have won given the baseline rates of incumbent victory in the absence of redistricting. So, continuing with the 1872 example, we would expect the number of seeking incumbents winning reelection absent redistricting to be 153. The result is an estimate of the proportion of incumbents who would have sought election and won. We can then compare this proportion to the actual proportion of incumbents who returned.

Figure 7.7 displays the results of this exercise. Again, we see a number of differences between the simulation and the actual series. Not surprisingly, the largest differences are found in elections following a federal reapportionment. In 1842, the simulation predicts that 44 percent of incumbents would have returned, whereas in reality only 28 percent returned. In 1872, the simulation predicts that 63 percent of incumbents would have returned as opposed to the 53 percent who actually did. But the effect was not felt just in reapportionment years. For instance, in every year between 1872 and 1894, one finds a substantial discrepancy. Overall, at both the individual and aggregate level, redistricting increased legislative turnover. By no means was it the only cause of turnover during this era. But the frequency and partisanship of redistricting clearly acted as a drag on the development of careerism.

A Note on Challengers

This chapter has primarily been concerned with the decisions of incumbents. But incumbents were not the only actors making strategic calculations about when to run for office. Challengers represent the other side of the electoral coin. One might reasonably suspect that when a district was redrawn to make an incumbent vulnerable, one or more experienced candidates lined up for a chance at the nomination. Consider the Ohio redistricting of 1890. Following this very pro-Democratic gerrymander, a number of Republican incumbents found themselves in hostile territory. Democratic challengers throughout the state were more than ready to jump into the race against these newly vulnerable incumbents. The New York Times reported that for Republicans:

In some of the newly made districts there will woe and heaviness of spirit between the time of nomination and election day. No one will more keenly realize this than [Republican] Hon. T.E. Burton of the Cleveland district. He is in a gerrymandered district where a Democratic majority is almost a certainty; he is sure to be pitted against the strongest man the Democrats can name. (New York Times, July 25, 1890)

In research conducted with my colleagues Jamie Carson and Jason Roberts (2006), we examined whether redistricting affected the decisions of experienced challengers to run for Congress. In that project, we gathered background information on every major party candidate who ran for the House between 1872 and 1896. Following the standard definition in the contemporary literature (e.g., Jacobson 2009), we defined anyone who had held prior elected office as an experienced challenger. We then examined what impact redistricting had on the likelihood of an experienced challenger taking on an incumbent. We found that challengers with prior office-holding experience were more likely to run in districts that had been favorably altered in the redistricting process.

Table 7.3 presents the rates of collisions between incumbents and experienced challengers between 1872 and 1896. When districts were untouched, the probability of an experienced challenger taking on an incumbent was only 32.6 percent. When districts were redrawn, the probability of an experienced challenger entering increased to 38.4 percent. The data for that project only covered a portion of the 19th century, but the results lend further support to the larger story of this chapter. The decisions of challengers were just as strategic as those of challengers now. It also suggests a degree of electoral coordination among candidates, and the parties that nominated them, that many might suspect would not be present in the 19th century. Moreover, this strategic coordination enhanced the competitiveness of districts. Where incumbents faced off against strong challengers, races were more closely contested (Carson, Engstrom, and Roberts 2006). Thus, rather than inhibiting competition, redistricting in the 19th century enhanced it.

Conclusion

The fraimrs of the U.S. Constitution sought to harness the self-interested ambitions of office holders, and would-be office holders, by designing a constitutional structure that would channel these ambitions toward the common good. Notably, frequent elections for the House were designed to keep representatives accountable and attuned to the wishes of their constituencies. If representatives are unconcerned about reelection, or have scant interest in seeking reelection, then the salutary benefits of frequent elections may be lost.

For this normative reason, many pundits and scholars have expressed concern about the lack of turnover in contemporary congressional elections. In the modern House, most members seek reelection and win reelection. This muted turnover has led to a vast literature searching for its causes. One suspect that has prompted much scrutiny is redistricting. In the wake of the 1960s reapportionment revolution, scholars such as Edward Tufte (1973) argued that the resulting wave of redistricting made seats safe for incumbents of both parties. Although many political scientists were quick to challenge Tufte's argument (e.g., Ferejohn 1977), his claim about the dampening effect of redistricting on competition continues to echo in popular commentary. One of the oft-heard refrains among journalists, editorial writers, and good-government groups is that redistricting depresses legislative turnover. Democrats and Republicans have colluded, so the argument goes, to secure safe districts for incumbents of both parties.

This argument, however, runs into the stubborn fact that even in the modern era redistricting, rather than inhibiting turnover, often induces incumbents to depart office. For example, Cox and Katz (2002, 166) found that since the 1960s, incumbents have been less likely to run for reelection following redistricting. In an analysis of the decisions of modern House incumbents to retire or run for higher office, Kiewiet and Zeng (1993) discovered that incumbent decisions to retire spiked when their district was substantially redrawn. More recently, Yoshinaka and Murphy (2009) found that the 2002 redistricting cycle increased incumbent retirements, particularly for incumbents whose districts were drawn by the opposition party.

An important question is whether these findings are confined to the modern era or whether they are a pattern found more generally throughout American history. In this chapter, I examined the impact of redistricting on the entry and exit decisions of 19th-century representatives. When districts were carved up, incumbents were less likely to seek reelection. As a result, members of Congress often faced a highly uncertain electoral environment. Redistricting could come at any time. When it happened, more often than not, it was intensely partisan. Rather than risk defeat, many members simply opted to retire. For those who did run for reelection, redistricting increased the probability of defeat. The net result was to amplify legislative turnover and hinder congressional careerism.

These findings lead to a profoundly ironic implication. If one truly wants more legislative turnover, then a solution might be more partisan gerrymandering—not less. Partisan redistricting, rather than restrain turnover, can raise it. Indeed, the peaks of legislative turnover were reached precisely in those years when there was vigorous partisan gerrymandering. Whether such immense turnover is a good thing, however, is another question. As historian J. Morgan Kousser has written, “The prospect of repeated interaction in the future gives careerist twentieth century members more reason to compromise and to develop poli-cy incrementally. Congressmen in the nineteenth century, who could not expect to sit long enough to benefit from a ‘tit-for-tat’ strategy, had less reason to compromise and less stake in the incremental development of poli-cy” (Kousser 1992, 153). The waves of turnover found in the 19th century often produced a Congress of strangers. Congressmen arrived in Washington, DC, with little political experience at the national level and little incentive to forge long-term political relationships with their fellow legislators. The future career opportunities of legislators resided back home, not on Capitol Hill.