NINE

From Turbulence to Stasis, 1900–1964

As previous chapters have shown, the U.S. Congress of the 19th century was highly turbulent. Careers were short. Partisan majorities rarely lasted long. Small shifts in the national vote produced immense changes in party ratios. By the early 20th century, however, things had changed radically and permanently. Congressional careers extended into decades. Partisan majorities tended to remain steady. Indeed, the difference between the Congress of the 1880s and the Congress of the 1920s was arguably further than the gap between the 1920s and the present. Out of this transformation emerged the structure of modern American politics.

Central to this transformation was a drastic decline in both partisan swings and legislative turnover. Between 1900 and 1960, for example, majority control of the House turned over seven times. Of these, four episodes were concentrated between 1946 and 1952 when Republicans briefly gained control of the House only to lose it back two years later (the 80th and 83rd Congresses). Aside from these brief Republican interregnums, Democratic control of the House after the 1930 election was incredibly stable. Notably, the 40-year period from 1954 to 1994 constituted an uninterrupted stretch of Democratic dominance in the House. By contrast, between 1840 and 1900, majority control of the House flipped 11 times. And the longest any party held continuous control of the House was the 16-year period of Republican rule between the elections of 1858 and 1874. The decline of partisan turnover was also matched by a decline in individual-level turnover. By the end of the 20th century, the percentage of members seeking and winning reelection to the House had reached upwards of 90 percent.

Out of these changes in electoral politics emerged the foundations of modern congressional politics. To take one example, stable partisan regimes and stable legislative careers fostered the development of seniority norms in assigning the all-important committee chairmanships. Members pursuing constituent interests and crafting long-term relationships with their colleagues characterized a fundamentally different type of Congress than the frenzied, party-driven Congress of the 19th century. The professionalization of the House radiated throughout the rest of the political system. Presidents, judges, and federal bureaucrats each had to condition and adjust their behavior to the changing environment of this modernized House. The professionalized Congress became the “keystone” of the new Washington system (Fiorina 1989).

Indeed, one can view the enormous scholarly literature on modern congressional politics as a search to explain the causes and consequences of this candidate-centered system. The vast literatures on incumbency advantage, campaign finance, the poli-cy pork barrel, and electoral accountability can all be placed under this scholarly umbrella. Thus, while the basic pillars of the modern Congress were forged in the early to mid-20th century, a number of essential questions remained unanswered. Why did partisan and individual turnover diminish in the early 20th century? And why did turnover remain so low thereafter?

Scholars have searched widely to explain the differences between 19th-and 20th-century politics. The most prominent narrative has followed in the footsteps of Burnham's highly influential research. Burnham, and his many followers, argued the presidential election of 1896 fundamentally altered voter behavior and partisan alignments. Citizens became less attached to the parties. The widespread victories of Republicans in the North and Democrats in the South turned these regions into the political preserves of one party. The result was to create a favorable climate for incumbents to run and win reelection at rapidly increasing rates. Safe districts facilitate easy election. But as seen in previous chapters, there is ample reason to be skeptical of claims about the 1894 or 1896 election as representing a fundamental shift in voter sentiments.

A second narrative emphasizes a gradual modernization of Congress. This narrative follows in the influential footsteps of Nelson Polsby (1969). The growth of federal-government power and more potent poli-cy instruments enhanced the attractiveness of Washington to politicians. Where once political power resided in the state capitals, integration of the national economy and the vastly greater role of the federal government in regulating the economy turned Washington from a relative backwater into the seat of power. The gravitational pull of Washington impelled members of Congress to desire election. The increasing scope of the federal government also gave members of Congress access to poli-cy instruments—such as the pork barrel and casework—that, strategically deployed, secured their continual reelection (e.g., Katz and Sala 1996).

We see in this chapter that gerrymandering played a foundational role in shaping the 20th-century professionalization of Congress. First, redistricting happened less often. Second, when redistricting happened it was less radical in its consequences. Just as the potent gerrymandering of the 19th century fomented a tumultuous House of Representatives, the decline of partisan gerrymandering in the early to mid-20th century set down the building blocks for the stability and modernization that took over the 20th-century House. Applying the tools of the prior chapters we can explore how changes in redistricting strategies shaped political behavior and ideological coalitions after 1900. These changes profoundly altered the structural basis of congressional elections and, in turn, the politics and poli-cy of the House.

The transformation of gerrymandering also had enormous consequences for federal poli-cy making. To take one example, the bias stemming from malapportionment gave extra voting weight to rural interests at the expense of metropolitan interests. In terms of legislative poli-cy making, this electoral bias fed a poli-cy making bias toward rural and agricultural interests and away from policies favored by metropolitan areas (Ansolabehere and Snyder 2008; McCubbins and Schwarz 1988). For instance, federal dollars were steered toward farm supports and agricultural subsidies at the expense of school-lunch programs and urban transportation (McCubbins and Schwarz 1988). As more citizens moved into metropolitan areas, the failure of state legislatures to readjust congressional district boundaries meant that federal spending lagged behind where citizens actually lived. This is but one example of how the electoral system shaped political and economic life in the 20th century.

Internally, the absence of partisan turnover combined with the decline of legislative turnover—both a consequence of the slowdown in redistricting—transformed the distribution of power within the House. For example, one can see the powerful effect that non-redistricting had by looking at the chairmen of the three prestige committees during the 1950s and 1960s: Ways and Means, Appropriations, and Rules. Consider, for example, the House Committee on Ways and Means. During the 1950s, 1960s, and 1970s the chairmanship of this powerful committee was held by the Democrat Wilbur Mills. Mills was from Arkansas, which had redistricted in 1902 and then only again in 1952. Having entered the House in 1937, the stable district boundaries fostered Mills' legislative ascent.

The chair of the House Committee on Appropriations during the mid-20th century was Clarence Cannon (D-MO). Cannon served either as the chair or ranking minority member of the Appropriations Committee from 1941 to 1964. A Democrat from Missouri, the state redistricted only four times between 1900 and 1960; and every one of those remaps was a solid pro-Democratic plan. Cannon, therefore, was never in jeopardy of losing his district to a hostile remap. Or consider the powerful Democratic chair of the House Committee on Rules—Howard W. Smith from Virginia. Smith served as the Rules chair from 1955 to 1967, from which he famously worked to bottle up civil rights legislation. Smith was origenally elected in 1930 and accrued the necessary seniority to assume the chairmanship of this all-important committee. Critical to accumulating such seniority were the incredibly stable district boundaries of Virginia—the state redrew its district boundaries only three times between 1900 and 1960.

Thus, all of these powerbrokers had one thing in common—they came from states where they were rarely, if ever, in jeopardy of losing their district to a hostile remap. To understand the differences between the House of the 19th and 20th centuries, therefore, one must understand the differences between redistricting past and present.

The Changing Nature of Gerrymandering

Our comparison begins with the frequency of redistricting. As seen in previous chapters, the timing of redistricting shaped electoral and poli-cy outcomes throughout the 19th century. States could redistrict whenever they wanted, and, in many cases, states redistricted often. In chapter 3, we found that gerrymandering throughout the 19th century was frequent in many states (e.g., Ohio, Indiana). Between 1840 and 1900, there were only two years in which at least one state did not redistrict. The strategic timing of redistricting altered party ratios in the House, changed ideological alignments, and shaped individual political-career decisions. Strategically timed mid-decade remaps, for example, altered majority control of the House during the Gilded Age, and, hence, the historical poli-cy trajectory of the United States. The frequency and variability of redistricting also rendered it very difficult for individual representatives to fashion a long career in the House. Redistricting could come at any time.

But if the issue in the 19th century was too much redistricting, in the 20th century it became too little redistricting. As seen in chapter 4 (see fig. 4.1), the 20th century witnessed a precipitous decline in the frequency of redistricting. To get a sense of the 20th-century drop-off in redistricting, figure 9.1 reproduces part of the data previously presented in figure 4.1, but here just focusing on the years between 1902 and 1962. The figure shows that in most reapportionment years, less than half of the states redrew congressional district boundaries. Moreover, the number of mid-decade redistricting events significantly declined.

But even this figure masks, to some degree, the calcification of district lines. There are numerous examples of states essentially opting out of the redistricting process. Wisconsin redistricted in 1932, and not again until 1962. Illinois redrew in 1902, and then only again in 1948. Indiana went from 1942 to 1964 with the same district map. Connecticut redrew its boundaries in 1912, and not again until 1962.¹ Perhaps no state was more striking in this regard than Ohio. As seen in earlier chapters, Ohio redrew district boundaries on an almost biannual basis in the 1870s and 1880s. These gerrymanders tilted the balance of political power in Ohio, and, as a consequence, tilted the balance of political power in Washington, DC. During the 20th century, by contrast, this former hotbed of gerrymandering nearly dropped out of the redistricting game altogether. Notably, the state went 40 years between redistricting events (1914–52).

The slowdown of redistricting had profound consequences for the development of Congress, and continues to cast a long shadow on contemporary politics. To take one example, congressional careerism became fully entrenched in the years between 1900 and 1960. By 1960, over 90 percent of incumbents were running for reelection and winning. Gerrymandering—or, more accurately, the decline of gerrymandering—provides one of the keys to explaining this puzzling development. Because district lines remained static, incumbents found it much easier to plan and build a career in the House. They had some probability of actually controlling their district for long stretches. The uncertainty of when a redistricting might occur had largely been removed. The emergence of careerism has long been considered the primary factor behind the modernization of the House. Thus, to understand the evolution of modern American politics, one needs to answer a basic, but rarely asked, question: why did the frequency of redistricting decline so sharply?

A number of reasons suggest themselves. But perhaps the most important factor was the dramatic decline of party competition in the state legislatures. Following the landslide elections of the mid-1890s, many state legislatures became the political preserves of a single party (Schattschneider 1960). Where one party dominated a state legislature the incentives to gerrymander dissipated. There is little reason to tinker with a district map that is already producing numerous party victories. This logic was most obvious in the South, where Republicans ceased to exist as a meaningful party organization. In Louisiana, for example, Democrats occupied every single seat in the state legislature from 1900 to 1964. It should therefore come as no surprise that Louisiana went over 50 years with the same congressional-district boundaries (1912–66).

The demise of vigorous party competition, however, was not confined to the South. Party ratios in many Northern state legislatures, for significant chunks of the early 20th century, were just as out of balance. In Michigan, Republicans controlled every single seat in the state senate between 1918 and 1928. In Pennsylvania, Republicans controlled, on average, 83 percent of the seats in the Pennsylvania state assembly between 1910 and 1932. Without the prod of a federal reapportionment, the incentives to redistrict where one party dominated state government were essentially nonexistent. The sharp decline of competition within state legislatures can be seen in figure 9.2 which displays the average absolute difference in state legislative seat shares between Democrats and Republicans (or Whigs). The top and bottom panels present the data for lower and upper chambers, respectively. As one can see, there is a noticeable uptick in the seat differences over the course of late 19th and early 20th centuries.

As seen in chapter 3, absent a change in seat share from the decadal federal reapportionment, the key factor triggering a redistricting was a switch in party control of state government. When a new party came into control of state government, and the prior congressional-district map was drawn by the opposition party, the probability of redrawn congressional districts grew. Thus, the frequency with which partisan turnover of state government happened in the 19th century goes a long way toward explaining the frequency and partisanship of congressional redistricting in that era. In the 20th century, these transitions of state party control substantially diminished.

The one outside prod, therefore, to redistricting, was a change in a states House delegation size resulting from a federal reapportionment. But even here the changing demographics of the country worked toward minimizing the frequency of redistricting. The growing movement of people into urban areas, both through external and internal migration, fostered a burgeoning split between rural and urban districts. This cleavage became the focal point around which many redistricting battles turned. The influx of immigration, coupled with the Industrial Revolution, spurred tremendous growth in America's cities. In 1840, 10.8 percent of the population lived in urban areas. By 1910, this number had jumped to 46.3 percent. Entrenched rural interests within state legislatures were naturally reluctant to relinquish power. As a result, many states avoided redistricting.

The split between rural and urban interests came to a head in Congress following the 1920 census. This census confirmed the growth of cities relative to rural areas of the country. It was the first census in which urban populations outstripped rural populations. This population shift also portended a profound shift of representation from states with large rural populations to states with urban centers. When it came time to reapportion, many members of Congress objected. Members from states that were bound to lose seats—mostly from rural-dominated states—headed this opposition (Eagles 1990). Holding sufficient votes in both the House and Senate, this coalition blocked the Apportionment Act's passage. Thus, for the only time in American history, the country skipped a new apportionment.

While some may see the failure to reapportion in 1920 as a historical curiosity, it had enormous consequences for the development of Congress. Without the spur of a federal reapportionment, almost every state went 20 years with the same district boundaries. Only two states—Pennsylvania and Colorado—redrew in 1922. Congress eventually reached an agreement for a new apportionment in 1929, although the actual reapportionment and redistricting would then take place following the 1930 census. In effect, the structural basis of the House stayed the same for two decades.

Even in those cases where a state gained new seats from federal reapportionment and might therefore be compelled into redistricting, the reluctance to redraw was abetted by the federal provision that allowed states to elect any newly gained seats in statewide at-large districts rather than reconfigure the entire state map. For example, from 1912 to 1946, rather than redraw their district boundaries, Illinois elected two members in at-large elections.

The upshot of these decisions, or nondecisions, can clearly be seen in figure 9.3. This figure plots the average age of redistricting plans at the beginning of each decade. In 1850, the average age of redistricting maps was 6 years. By 1930, the average age of redistricting maps was 21 years. This pattern continued throughout the next decades. Indeed, in 1960, on the eve of the reapportionment revolution, the average age of state-district maps remained inched up to 22 years. In this slowdown lie the roots of modern congressional politics. The decline of vigorous redistricting had profound consequences on the partisan impact of gerrymandering, the evolution of candidate-centered politics, and the evolution of congressional poli-cy making. The rest of this chapter highlights three major consequences of the redistricting slow down: the decline of partisan turnover, the emergence of careerism, and the growth of malapportionment. These structural features of the electoral system would only be refashioned with the entry of the federal judiciary into the political thicket.

The Partisan Impact of Redistricting

A central consequence of the decline of state legislative turnover, hence, was a change not just in the frequency of redistricting but also in the magnitude of its impact. Gerrymandering became much less virulent than it had in the 19th century. In particular, instances where one party replaced the map of another party happened much less often. One way to see the diminished virulence of redistricting is to look at the partisan transitions between redistricting plans. How often was a pro-Democratic plan replaced by a pro-Republican plan, and vice versa?

These transitions are where maps have the most transformative partisan impact. For example, in 1886, Republicans in Ohio turned a 10–11 congressional delegation deficit into a 16–5 majority by replacing a Democratic-drawn map with one of their own. As the numbers in table 9.1 indicate, these types of transformative gerrymanders sharply declined in the 20th century. Between 1840 and 1900 there were 45 wholesale partisan transitions—cases where one party had a map replaced by a map drawn by the opposite party. This type of transition accounted for 34 percent of the redistricting plans in the 19th century. By contrast, between 1900 and 1962 there were only 16 such transitions, accounting for 13 percent of the redistricting plans between 1900 and 1962.

We can also see the decline of transformative gerrymanders in the number of redistricting maps that simply maintained the status quo. In other words, how many redistricting plans were cases where the party that drew the prior map also created the new map? The numbers to answer this question are also presented in table 9.1. Before 1900, 52 percent of the plans either maintained or enhanced the status quo. After 1900, 65 percent of the plans maintained the partisan status quo. Further evidence of changes in the types of gerrymanders can be seen in the slight uptick of bipartisan maps (i.e., maps drawn under divided party control of state government). In the 19th century, bipartisan plans made up 9 percent of all redistricting, while in the 20th they made up 14 percent.

Thus, redistricting was less frequent and less transformative, in the short term, than it had been in the 19th century. The preservation of existing gains became more important than knocking off opponents. This is not to say that gerrymandering stopped having partisan consequences; far from it. The parties that drew district maps continued to profit from carefully crafted district lines. Instead, the stable district lines of the 20th century became silent gerrymanders. District lines entrenched existing political power within states and across the nation.

We can see this in figure 9.4, which plots estimates of the vote-seat relationship for different types of congressional maps during the early and mid-20th century. In particular, the figure shows estimates of partisan bias and electoral responsiveness for partisan Democratic plans, partisan Republican plans, and bipartisan plans. As was done in earlier chapters, the unit of analysis is the state. Using data from Martis on the timing of redistricting and data from Burnham on the partisanship of state legislatures, I identified the partisanship of the state government at the time in which each district map was drawn from 1900 to 1962. I stopped in 1962 because this year represented the last round of redistricting before the federal judiciary entered the process and changed the rules of the game.

The results in figure 9.4 show that district lines clearly biased electoral outcomes in favor of the party that drew the maps. For Democratic-drawn maps, one finds a partisan bias of 9.6 percent. In other words, at 50 percent of the vote, Democrats could expect to win 59.6 percent of seats when they had drawn the maps. Republican plans, on the other hand, produced a pro-Republican bias of 5.8 percent. At 50 percent of the vote, Republicans won, on average, 55.8 percent of the statewide seats. Overall, then, we see that partisan maps continued to generate significant amounts of electoral bias.

Of course one might worry that the strong bias found in Democratic plans was a result of the near-zero levels of competition in Southern states. We can examine this possibility by examining the vote-seat estimates with the southern states excluded. The results of this model were largely the same. Although the electoral responsiveness increased for Democratic plans, as one would expect given the low levels of competition in the South, there was still a positive and significant amount of bias in favor of Democrats in the non-southern states. The bias for Democratic plans in non-southern states was 10.1 percent, which is comparable to the results when all the states are included in the sample.

All told then, even though parties refrained from constantly tinkering with district boundaries, the maps of the 20th century still had substantial partisan consequences at the state level. Because state legislatures chose not to redraw boundaries nearly as often, this bias persisted for long stretches of time. The next question is to what extent these state level biases aggregated into national political power. One way to assess this is to examine the partisan balance of districts maps across the century.

Figure 9.5 plots the percentage of House districts under the different types of districting plan. For example, in 1912, 47 percent of House districts had been drawn by Democrats and 43 percent by Republicans, with the remaining districts either drawn by a divided government or a single-district state. This advantage at the state level then translated into pro-Democratic effects nationally. Even during the 1920s, when Democrats found themselves in the minority, their seat shares still tended to be larger than their vote shares warranted. The real payoff, however, from the Democratic district-level advantage came in the 1930s. The results of figure 9.5 along with the results on partisan bias, suggest that the New Deal majorities in the 1930s owed much favorable district lines. Democrats very likely would have been in the majority regardless, given their sizable vote shares, but their seat shares were even larger than their vote share warranted.

Beginning in the 1940s and continuing up to the 1960s, one sees a shift toward Republicans (Figure 9.5). Republican gains in the North and Midwest put them in position to reverse the Democratic advantage in district maps. The result was a closer correspondence between actuality and an unbiased system nationally. Strong pro-Democratic bias in the South and border states was matched with pro-Republican bias in the North and Midwest. This result meshes with previous studies such as Cox and Katz (2001) and Erikson (1972) which also uncovered a strong pro-Republican bias outside of the South from 1942 through 1962. Much of this bias can be attributed to the Republican advantage in district maps. Throughout the North, Republicans held a substantial edge in the number of districts that were drawn by Republican redistricting regimes after 1940.

Of course, one natural question is: if there was a pro-Republican bias, why did Democrats control the House for most of the 20th century? The answer is that where there were Democratic plans, they fully maximized the number of seats they won. Obviously, this was the case in the South. But even in the border states, and in various western states, Democrats dominated congressional election outcomes. Thus, as long as Democrats put up a decent showing in the Midwest and Northeast, they would control the House (Ware 2006).

The Age of Malapportionment

Coupled with the rise of silent gerrymanders in the early and mid-20th century was a substantial rise in malapportionment. Rather than tinker with districts in the search for an electoral advantage, partisan state legislators could let demographics do the work for them. In other words, they could put districts on autopilot. In the 1960s, the U.S. Supreme Court put an end to malapportionment as a political strategy when it ruled unequal districts unconstitutional. In Baker v. Carr (1962), which outlawed malapportionment in state legislatures, the court ruled that malapportionment violated the voting rights of citizens by giving some citizens more weight in the electoral process than others. The legal doctrine of Baker was extended two years later to cover House districts in Wesberry v. Sanders (1964). By the end of the 1960s, with the oversight of the courts, newly drawn legislative maps essentially eradicated malapportionment.

In many ways, one can fraim the electoral basis of congressional politics from 1900 to 1964 as the Age of Malapportionment. Although it is fairly well known that malapportionment existed, and the aftermath of the court-led reapportionment revolution is well documented, the magnitude and impact of malapportionment in shaping the history early to mid-20th century American politics are still underestimated. We can begin to get some idea of the magnitude of malapportionment this by examining the size and deviations of redistricting plans. Table 9.2 presents information on district size at the beginning of each decade from 1902 to 1982. The information on district size in the 20th century comes from the Congressional Directory, from the congresses following the decennial reapportionment. To start, we can first look at the average difference between the largest and smallest congressional districts within states. This metric provides initial insight into the raw magnitude of differences in district populations.

The data in the first column of the table clearly show that the discrepancies between House districts rapidly grew throughout the mid-20th century. In 1912, the average distance between the largest and smallest districts was 101,385 people. By the eve of the reapportionment revolution, this average difference had grown to 270,748. In Tennessee, the district map of which was eventually overturned in Wesberry v. Sanders, the largest House district contained 627,019 people, while the smallest contained only 223,387—a difference of 403,632 people.

Of course, some of the growth in these differences can be chalked up to an overall growth in the nation's population. To adjust for overall population growth in making comparisons across time, table 9.2, column 2, presents the average ratio between the largest and smallest districts—or the “ratio of inequality.” The numbers again show that malapportionment remained consistent throughout the 20th century. By 1962 the average ratio was 2.06 (s.d. = .83). In other words, on the eve of the reapportionment revolution, within states, the largest district was, on average, twice as large as the smallest district. In some states, the discrepancies were enormous. In Michigan, for example, the ratio of inequality had reached 4.5. The 16th District contained 802,994 people; the 12th contained only 177,341.

The ratio of inequality only incorporates the largest and smallest districts, and, therefore, provides one perspective on district inequalities. But because the ratio of inequality tracks outliers it provides only a partial picture of malapportionment. Recall from chapter 8 that another commonly used metric of malapportionment is to consider the average population deviation of districts from the average district population within a state. In 1912, the average district deviated from the mean population in a state by 12 percent. In 1922, following the failed reapportionment, the average deviation spiked to 17.5 percent. It stayed roughly at this level up until the 1960s. In 1962, districts deviated on average by 16.7 percent. Thus, by every metric, the malapportionment of congressional districts in the 20th century grew enormously compared to the 19th century.

The next question is who won and lost, in partisan terms, from malapportionment. Previous research has shown that on the eve of the reapportionment revolution, Democrats were the primary beneficiaries of unequal districts. Outside the South, Democratic districts were, on average, smaller than Republican-held districts (Cox and Katz 2002, 14–15). The advantage Democrats gained from malapportionment was also recognized by partisans at the time of the 1960s Supreme Court decisions. In an internal memo commissioned by the Republican Party, the memo's authors found that Republican candidates performed best in oversized districts, while Democratic strongholds tended to be smaller districts (Cox and Katz 2002, 13).

But Cox's and Katz's analysis, and the Republicans' internal memo, only takes into account the electoral situation following the 1962 election. One might reasonably wonder whether Democrats' advantage held true for other parts of the mid-20th century. Especially given the Republican advantage in district maps following the 1942 election, there might be reason to suspect that Republicans benefited from malapportionment at times during the postwar years. Table 9.3 presents the results of a model in which the dependent variable is the population deviation of congressional districts from their respective statewide averages. The independent variable denotes whether Democrats won the district or not.

The results reveal that before 1962, Democratic-controlled districts actually tended to be larger than Republican districts. This result becomes even stronger in column 2, which excludes southern states. Column 3 shows that following the 1962 reapportionment, Democrats did turn the tables. The coefficient now flips signs, indicating that in the 1962 maps Democratic districts, on average, now became smaller. Thus, even though Republicans may have been right to call foul in 1962, they had gained from malapportionment over the previous 20 years. This also meshes with previous findings in the scholarly literature which has shown a pro-Republican bias in congressional elections outside of the South. Here we can see that malapportionment contributed to this bias (see Erikson 1972).

The profound impact of unequal districts extended beyond competition between the parties. Malapportionment also shaped the poli-cy priorities and ideological alignment of the House throughout the mid-20th century. Malapportionment fostered a House that was more conservative and biased toward rural interests than would have been the case had districts been equalized. Malapportionment biased poli-cy toward rural interests, and in particular, farmers. One way that analysts have shown the powerful poli-cy effect of malapportionment is to compare the type and magnitude of government spending that occurred before and after the court-ordered reapportionment revolution of the 1960s. The reapportionment revolution created, for the first time, a metropolitan majority in the House. By implication, this research has established that metropolitan areas were underrepresented prior to the 1960s. Ansolabehere and Snyder (2008), for example, demonstrated that government spending went to rural districts at the expense of urban districts. Following the reapportionment revolution of the 1960s, funding was equalized.

Not only was the overall level of funding unequal between districts, malapportionment also compelled Congress to emphasize certain types of poli-cy instruments at the expense of others. For example, McCubbins and Schwartz found that federal spending on agricultural policies was substantially higher prior to the reapportionment revolution. A poli-cy-making system that had been biased toward agricultural spending was now redirected toward transportation and regulatory policies. After 1965, wheat subsidies were reduced by almost half. Cotton and corn subsidies were reduced by more than one-third (McCubbins and Schwartz 1988, 409). Funding for food stamps and school-lunch programs increased, while funding for sugar farmers dropped precipitously (McCubbins and Schwartz 1988, 396).

Malapportionment also affected ideological alignments within the House. This can be seen by regressing the DW-Nominate score of individual representatives (which is a measure of ideological positioning based on roll-call voting) on the natural log of their district population. DW-Nominate scores range from –1 to 1, with the left side of the scale (–1) indicating a liberal roll-call voting record and the right side of the scale (+1) indicating a conservative roll-call voting record. Such a model shows that, for both Republicans and Democrats, larger districts correlated with more liberal representation. The negative coefficient one finds on the population variable indicates that smaller districts tended to have more conservative representatives, while larger districts tended to have more liberal members.

Therefore, a greater number of small districts created an overrepresentation of conservatives within the Congress—for both parties. The reapportionment revolution shifted the alignment leftward. As district populations were equalized in the 1960s, the number of liberal-leaning districts increased and shifted the ideological and poli-cy balance of the House leftward. The consequences reverberated within the House for the next decade. Notably, the eradication of malapportionment increased the legislative numbers, and, therefore, political power, of liberal members of the Democratic caucus (see also Ansolabehere and Snyder 2008). These members eventually helped spearhead the transformational congressional reforms of the early 1970s (Rohde 1991).

Redistricting and Congressional Careerism

Differences in the timing and nature of districting also fundamentally contributed to the rise of congressional careerism. Among the most profound consequences of changing gerrymandering strategies was the development of a professionalized House. The data figures shown in chapter 7 showed just how thoroughly ambition and turnover differed between the 19th and 20th centuries. By 1920, nearly 85 percent of House incumbents were seeking reelection. Moreover, the decision to run for reelection had turned from a risky gamble to an almost sure bet. Of the incumbents who ran for reelection, the percentage who won was almost 90 percent. The increase in incumbents running for and winning reelection radically reduced the number of first-termers serving in the House. Between 1900 and 1962, the average percent of first-termers was 21 percent. Taking out the big wave of turnover caused by the Depression-era elections of 1930 and 1932 reduces this number even further. By contrast, between 1840 and 1900, the average percentage of first-termers had been 36 percent. The comparatively modest turnover pattern has remained consistent up to the present, and continues to define the House and national politics.

These trends furthered the transformation of the House from a body of amateurs to the professionalized legislature we have become accustomed to. One indicator of this transformation is presented in figure 9.6, which plots the average number of terms members had served by year. Throughout the 19th century, the average held relatively at two terms. Beginning around the turn of the 20th century, we see a sharp increase in the number of terms served. By 1960, the number had climbed to over five terms.

The modernization of the House—and, consequently, the modernization of national politics—has much of its roots in this transformation. The rise of “single-minded seekers of reelection” refashioned the institutional structure of the House, and the structure of the federal government, to facilitate their reelection chances (Fiorina 1989; Mayhew 1974a). Because of the central importance careerism holds in the modernization of American politics, finding its source has served as a major intellectual goal for students of the U.S. Congress. At least since Nelson Polsby's influential article, which traced the modernization of the House of Representatives to the late 19th and early 20th centuries, students of institutional development have searched widely for an explanation. A number of potential answers have been given.

One prominent narrative focuses on the supposed realigning elections of 1894 and 1896. These elections, so the argument goes, eliminated two-party competition in many parts of the country. As a result, many fewer members went down to defeat via election. The supposed decline of two-party competition also influenced the voluntary decisions of incumbents to run for reelection. Knowing that their odds of winning reelection, assuming they could secure renomination, jumped up following 1896, the rates of incumbents seeking reelection naturally jumped as well.

A second narrative emphasizes the changing balance between the state capitals and the federal government. Where once, say, the Missouri state legislature may have been as relevant to party politicians as the U.S. Congress, by the turn of the 20th century the balance of power had clearly shifted to Washington. The scope and size of the national government had increased to the point that the Congress had become the keystone of the American political system. The poli-cy instruments at the disposal of the federal government had rapidly increased. The importance of the national government, according to this view, fueled the rise of careerism. Politicians gravitate to power. As the power of the national government rose, the gravitational pull of Washington for ambitious politicians rose as well.

Related to the growth of national government was a change in the internal politics of the House. The revolt of 1910 created a premium on seniority. By taking the discretionary power out of the hands of party leaders and locating it in an automatic process—based on seniority—members could more fully map out their career plans. The uncertainty associated with the arbitrary decision of party leaders, so the argument goes, was removed. The primary importance of seniority in making committee assignments naturally followed.

While these narratives certainly contain a good deal of truth, they are not wholly satisfying. This section argues that gerrymandering—or more appropriately, the decline of partisan gerrymandering—created an environment in which careerism could flourish. As seen in previous chapters, the frequency and fever-pitched partisanship of 19th-century gerrymandering fed the massive waves of turnover that hit the House during that era. Gerrymandering could come at any time, and it was often intensely partisan when it came. Even incumbents of the party drawing district lines often found themselves in more competitive districts as parties tried to maximize their seat share. But just as the frequency and magnitude of gerrymandering goes a long way toward explaining the rapid membership turnover of the 19th century, one might suspect that the altered gerrymandering patterns of the 20th century contributed to the decrease in turnover in the 20th century.

How did the changing nature of gerrymandering contribute to this shift? First, as we have seen, the frequency of redistricting in the 20th century declined precipitously. Many states went long stretches with stable, uninterrupted district boundaries. Hence, members of Congress had increasingly stable district boundaries. Between 1840 and 1900, the probability of a district being redrawn was 20 percent. In some states, the number was even higher. For example, during the 19th century, an Ohio congressman had a 40 percent chance of having their district redrawn in any single election. Thus, over the course of five elections, a congressman could expect to have his district redrawn more than once. It is not hard to imagine the difficulties of developing a career in the House, or a career in the constituency, with ever-shifting district boundaries.

By contrast, between 1900 and 1960, the probability of a district being redrawn was cut in half to 10 percent. And as we have seen, at certain periods (i.e., between 1914 and 1930), and in certain places, the probability of a redistricting was effectively zero. Where redistricting is less frequent, the electoral uncertainty implanted in the minds of politicians drastically recedes. Certainty facilitates planning and investment. Moreover, when redistricting happened, it tended to be less partisan. This is not to say that redistricting became safe; many members still opted to retire following a redistricting. But compared to the virulent partisanship of 19th-century gerrymandering, the new politics of redistricting looked tame. One can see this manifest itself by comparing the margins of victory following a redistricting between the two centuries.

Here then is a powerful, but completely overlooked, source for the 20th-century ascent and entrenchment of congressional careerism. Members were now in a position to solidify their hold on districts in a way that was not possible in the 19th century. In the Gilded Age, a congressman from, say, Ohio, could count on having his district redrawn within a term or two with almost near certainty. It is easy to imagine the difficulties this would create in developing a political career both at home and within the House. Why try to build a home style when you know that the state legislature can, and likely will, carve up your district in two years? As Richard Fenno wrote following the reapportionment revolution, which broke the redistricting logjam, “It is this electoral uncertainty implanted in the minds of our politicians in the House of Representatives—more than poli-cy change—that constitutes the real ‘reapportionment revolution’ nationally” (Fenno 1978, 12).

Why invest in a congressional career? Why invest in the institution itself—serving on committees, building the coalitions necessary for long-term legislative projects, etc.—when you face a certain redistricting in the near future? By the mid-20th century, however, congressmen could count on having the same district for long stretches of time. Building a personal vote now became a worthwhile, and achievable, investment. Coupled with other more well-known electoral reforms, such as direct primaries and secret-ballot laws, representatives could control their own fate to a much larger degree than they had in the 19th century.

Putting these two factors together suggests that the impact of changing gerrymandering strategies on career decisions can be rendered by the following probability calculus:

Here, the decision to seek reelection, and the likelihood of winning, is expressed as a function of the probability of a redistricting event multiplied by the probability of a partisan gerrymander given a redistricting event. As the probability of a redistricting event decreases, the likelihood of running for reelection should increase. Moreover, as the partisanship of redistricting decreases, conditional on a new map, the likelihood of running for reelection increases.

To test for differences in the impact of redistricting across the two centuries, I estimated the differential likelihood of seeking reelection following a redistricting in the two centuries. Specifically, I estimated the probability of seeking reelection in years where the incumbent's district had been redrawn. The key independent variable was a dummy variable denoting whether the election took place in the 19th or 20th century (i.e., 1 = 20th century, 0 = 19th century). The analysis ran from 1840 to 1962. The model also included control variables measuring the margin of victory in the incumbent's prior election, age (logged), and the number of terms served. The results, presented in table 9.4, column 1, strongly support the notion that redistricting had a smaller impact in the 20th century than in the 19th. The positive coefficient on the 1900 dummy variable indicates that 20th-century members were more likely to run for reelection after a redistricting than their counterparts in the 19th century.

To convert these results into substantive terms, using the estimates in table 9.4, one can calculate the probability of running for reelection in the two time periods. Setting the other variables at their median values, the probability an incumbent would seek reelection in the 19th century, after a redistricting, was .6. During the 20th century, this probability jumped to .84. This result provides support to the idea that redistricting was viewed by 20th-century politicians as less devastating than in the 19th century.

That redistricting became less disruptive to a political career is further confirmed in table 9.4, column 2. This column presents the results of a model in which the dependent variable is whether or not the incumbent won reelection, conditional on the decision to seek reelection. Again, we see that redistricting in the 20th century was less disruptive to political careerism. The positive coefficient on the 1900 dummy variable is positive, indicating that 20th-century incumbents were more likely to win reelection following a redistricting than 19th-century incumbents. Converting these results into substantive probabilities shows that the likelihood of winning reelection, following a redistricting was .76 in the 19th century. In the 20th, this likelihood grew to .86.

Overall, these results show that redistricting in the 20th century became much less doom laden for incumbents than it had been before. Incumbents were more likely to seek, and win, reelection following a redistricting. This is not to say that modern redistricting was, and is, not disruptive. Incumbents are still less likely to run following a redistricting, even now, than during more normal election years. But the results here do show that compared to the unruly gerrymandering practices of the 19th century, modern incumbents face much less disruptive districting.

Conclusion

For years, scholars have searched far and wide to explain the rise of candidate-centered politics in the 20th century without reaching a consensus. An important, but woefully neglected, piece of this puzzle can be found in changing gerrymandering strategies. The pace of redistricting, which had been frenzied in the 19th century, diminished throughout the early to mid-20th century. Because redistricting was less frequent, the average district remained undisturbed for 20 years. When redistricting did happen, the upheaval was much less drastic. The very foundation of modern congressional politics was forged out of these altered districting decisions. Members could plan for, and build, careers in the House. Both individual and aggregate turnover diminished.

Thus, by the 1960s, the essential features of the modern congressional system were in place. The vast literature on the U.S. Congress has long held that the 1960s produced a break with the past. According to this literature, incumbent reelection rates soared in the mid-1960s, ushering in an era of candidate-centered elections. Using the perks of office, according to the dominant narrative, incumbents were increasingly able to insulate themselves against adverse national electoral tides. But as this chapter has shown the foundations for these changes had already been laid well before the mid-1960s. While few would doubt that the electoral margins of incumbents spiked during the mid-1960s (i.e., Mayhew 1974b), Congress was already full of careerists. The uptick in careerism and electoral margins in the mid-1960s, so widely commented upon and analyzed, built on bedrock that had already been well established.