This study of voting in presidential elections in 12 central cities from 1920 to 1996 shows that cities played a crucial role in the New Deal realignment that dominated presidential elections from 1932 to the 1960s. Since then, cities have declined as a share of the total electorate, but they still provide crucial votes for successful Democratic presidential candidates. As cities have increasingly deviated from national voting trends, however, their turnout rates have increasingly fallen behind the national rates. A call is issued for researchers to break down the suburban vote and to examine contextual effects on voting behavior.

Many scholars have documentedthe rise and fall of the urban electorate in national elections (Degler 1964; Lubell 1965; Sundquist 1973; Andersen 1979; Mollenkopf 1983; Edsall and Edsall 1991). The rise of the political for- tunes of cities followed the migration from farms to cities that, by the middle of the twentieth century, made cities the population centers of the country. The Democratic Party relied heavily on urban votes to build the New Deal coalition that dominated national elections from the 1930s to the 1960s. Similarly, the migration outward from cities to suburbs, which accelerated in the 1950s, fueled the rise to power of the Republican Party in national elec- tions beginning in 1968 and signaled the progressive marginalization of urban electorates in national politics.

Accounts of the urban electorate in presidential elections rely heavily on the theory of electoral realignment.1 According to realignment theory, approximately once in a generation (1860s, 1890s, 1930s), the existing party

at Glasgow University Library on October 15, 2012uar.sagepub.comDownloaded from

http://uar.sagepub.com/

system reaches a crisis when “emergent tensions in society . . . not adequately controlled by the organization or outputs of party politics as usual, escalate to a flash point” (Burnham 1970, 10). Political entrepreneurs take advantage of the crisis with new appeals, ideological differences widen, the coalitions behind the parties are shuffled, and a new party system is formed with a new dominant coalition. The realigned party system incorporates political demands that had been ignored by the old party system.

Franklin Roosevelt’s victory in 1932 is viewed as a quintessential critical election. Roosevelt was able to swing urban ethnics, blacks, and Jews into the Democratic Party, forming the nucleus, along with the “solid South,” of the Democratic coalition that dominated presidential elections for the next gen- eration.2 The flight to the suburbs, which accelerated after World War II, weakened the urban base of the Democratic Party. With suburban voters per- ceiving that liberal Democratic policies siphoned off their tax dollars to expensive and wasteful programs targeted on central cities and blacks, Republicans successfully used “wedge issues” to split off elements of the New Deal coalition (Wilson 1987; Edsall and Edsall 1991). By the 1992 elec- tion, Schneider (1991, 1992) estimated that suburbanites cast an absolute majority of the votes, and he proclaimed “the dawn of the suburban era in American politics.”

The above account of the rise and decline of urban electorates provides the backdrop for Bill Clinton’s vaunted “suburban strategy” that supposedly won him the 1992 and 1996 elections. According to Clinton pollster Stanley Greenberg, the “forgotten middle class” of white suburbanites, as exempli- fied in Greenberg’s (1995) research on Macomb County, Michigan, had abandoned the Democratic Party in the face of what they saw as its preoccu- pation with urban blacks. The key to Democratic presidential victories was to challenge the Republican Party for suburban votes, Greenberg argued, entic- ing back into the fold the so-called Reagan Democrats (Greenberg 1995).

To attract suburban votes, Clinton avoided identification with policies tar- geted to cities or to minorities, emphasizing instead policies that benefit peo- ple wherever they live. Clinton’s well-publicized attack on Sister Souljah in the 1992 campaign distanced him from Jesse Jackson and the liberal wing of the Democratic Party centered in cities. The highly successful Clinton/Gore bus tours avoided the inner cities and were resplendent with small-town, even rural, imagery. Other than enterprise zones, Clinton proposed few policies targeted on cities. Instead, he proposed broad-based policies, such as national health insurance and investments in education, that would benefit cities and suburbs, blacks and whites.3 Clinton’s two victories were widely credited to his ability to contest Republicans for the swing votes in the suburbs. Even

Sauerzopf, Swanstrom / THE URBAN ELECTORATE 73

at Glasgow University Library on October 15, 2012uar.sagepub.comDownloaded from

http://uar.sagepub.com/

though the conventional account, described earlier, of the rise and decline of the urban electorate is widely accepted, there is relatively little published analysis of city voting trends over time to back it up.4 The main reason is that voting statistics in the United States are reported by county, not by city.5

Unless the city boundaries correspond to the county boundaries, city election returns are not widely available. Before 1950, county voting returns usually corresponded closely to city returns, but with massive suburbanization over time, county returns have increasingly deviated from city returns.

One exception to the dearth of urban voting studies is a 1949 article by Samuel Eldersveld, who painstakingly collected data on presidential elec- tions for 12 cities from 1920 to 1948. In this article, we update Eldersveld’s data for the presidential elections since 1948.6 Eldersveld chose all cities with populations more than 500,000 according to the 1940 census, eliminating Washington, D.C., and Buffalo. The 12 cities are the following: New York, Chicago, Philadelphia, Pittsburgh, Detroit, Cleveland, Baltimore, St. Louis, Boston, Milwaukee, San Francisco, and Los Angeles. This sample of cities is by no means representative of the nation as a whole. It is biased toward older northern cities, although it does include border cities such as Baltimore and St. Louis, as well as the West Coast cities of Los Angeles and San Francisco. Our main justification for choosing these cities is that it enables us to piggy- back on Eldersveld’s work and thus track city voting over a long time period. Moreover, the 10 states represented by these 12 cities still control 215 elec- toral votes and thus are intrinsically important. Admittedly, our work is only exploratory, and further research is needed on urban voting in key sunbelt states such as Florida and Texas.

We use this aggregate voting data to examine the voting behavior of central-city and non-central-city electorates over time and to suggest some revisions in the conventional accounts, both scholarly and in the mass media, of the rise and decline of urban electorates. Our data are limited, however, because they only allow us to show how the urban electorate voted, not why it voted the way it did. The striking voting patterns that we uncover could be the result of the characteristics of individual voters who live in cities (e.g., their race or class), or voting behavior could be due to a “contextual effect” of liv- ing within central-city municipal boundaries. Our data do not allow us to decide between individual and contextual effects. However, municipalities are important units within our federal system, with billions of dollars of fed- eral money targeted to them. Moreover, municipalities may be important units for shaping political identities. Therefore, in the conclusion, we call for further research on our hypothesis that contextual effects arising from central-city residence significantly affect voting behavior.

74 URBAN AFFAIRS REVIEW / September 1999

at Glasgow University Library on October 15, 2012uar.sagepub.comDownloaded from

http://uar.sagepub.com/

RISE OF THE URBAN ELECTORATE: NEW DEAL REALIGNMENT

Eldersveld’s (1949) main argument in his original article was that the tra- ditional way of understanding American presidential elections in terms of sectional divisions (e.g., the South vs. the North) was no longer adequate. Rather, this view needed to be supplemented by an appreciation of a powerful urban-rural division in the national electorate. That is, rather than looking at presidential elections as being won by a candidate’s appeal to base and key swing regions of the nation, Eldersveld argued that we should look at elec- tions as hanging on a candidate’s ability to appeal to cities. This was the case because the most powerful predictor of how individuals voted had clearly become whether they lived in urban areas. By 1949, to a great extent regard- less of region, small-town and rural area voters chose Republican candidates, and big cities voted Democratic. Eldersveld showed that 12 of the nation’s largest cities in 10 states consistently turned out victories for Roosevelt and Truman.

Figure 1 shows that the shift of the urban electorate to the Democratic Party that drove the New Deal realignment actually began in 1928 with the nomination of Al Smith, the first non-WASP ever nominated by a major American party. As Lubell (1965, 49) put it, “Before the Roosevelt Revolution there was an Al Smith Revolution.” A Catholic of immigrant background whose career was based on the New York Democratic machine, Tammany Hall, Smith’s campaign was replete with urban imagery. Especially impor- tant was Smith’s opposition to the Klu Klux Klan and prohibition, stands that were popular with urban ethnics. Key (1955) noted that the 1928 election had all the earmarks of a critical realigning election, with voter turnout rising 26.5% from the previous election and the coalitions behind the parties undergoing major change (see also MacRae and Meldrum 1960; Sundquist 1973, 177).7

If Smith began the transition of urban voters to the Democratic Party, clearly Roosevelt completed it in the 1930s. Having grown up in the country, Roosevelt was not partial to cities. Furthermore, the Democratic Party prior to 1932 was more closely identified with Jeffersonian localism than with an activist federal government. Roosevelt ran for office in 1932 promising to balance the budget. But once in office, Roosevelt recognized the need for fed- eral action and enacted a series of urban-targeted programs, “spending nearly $20 billion in new federal funds to put nearly one-fifth of the unemployed to work at transforming the urban infrastructure” (Mollenkopf 1983, 70). Andersen (1979) has shown how Roosevelt’s urban appeal mobilized non- voters and people voting for the first time into the New Deal coalition.

Sauerzopf, Swanstrom / THE URBAN ELECTORATE 75

at Glasgow University Library on October 15, 2012uar.sagepub.comDownloaded from

http://uar.sagepub.com/

Eldersveld (1949) argued that the pluralities the Democrats gained in cit- ies were crucial to their victories in the electoral college. He classified a city, or cities, as having “won” the state if the urban plurality was 50% or more of the statewide plurality.8 Clearly, the urban vote in these 12 cities was crucial to the Democratic victories (see Figure 2). If it were not for the hefty Demo- cratic pluralities of these major cities, Eldersveld maintained, the Democrats would have lost the presidency in 1940, 1944, and 1948.

THE DECLINE OF THE URBAN ELECTORATE: SUBURBANIZATION

As shown in Figure 3, the relative size of the city presidential vote reached its highest point in 1948. For our 12 cities, the city proportion of the total national vote reached an impressive 21.3% in 1948, subsequently falling to only 5.9% by 1996. In 1948, for example, New York City cast more than 50% of the statewide vote; in 1996, New York City represented only 32.1% of the

76 URBAN AFFAIRS REVIEW / September 1999

Figure 1: Sum of Sample City Pluralities, 1920-1996

at Glasgow University Library on October 15, 2012uar.sagepub.comDownloaded from

http://uar.sagepub.com/

state vote. Similarly, the city of Chicago fell from 46.5% of the Illinois vote in 1948 to only 20.9% in 1996 (see Table 1).

Notwithstanding the relative decline of the urban electorate, looking back at Figure 2, one can see that city voters have continued to play important roles in presidential elections. With the exception of Al Smith, probably no presi- dential candidate of a major political party was more identified with cities than John Kennedy. The Democratic Party platform in 1960 attacked the Republicans for having turned their backs on cities and contained specific planks to meet the problems of cities as governmental units, calling for the creation of a cabinet-level department to coordinate urban programs (Rourke 1965, 155). According to Eldersveld’s (1949) criterion, our 12 cities were responsible for swinging 146 electoral votes to Kennedy in a very close elec- tion. Kennedy’s massive plurality in Chicago was just enough to overcome the Republican advantage downstate (see Table 2). Republicans complained that voting fraud in Chicago swung Illinois to the Democrats.

Sauerzopf, Swanstrom / THE URBAN ELECTORATE 77

Figure 2: State Electoral College Votes Won by Sample City Pluralities, 1920-1996 NOTE: A sample city’s plurality is classified as having “won” its respective state’s electoral col- lege votes if the city’s plurality represents more than half of the state’s plurality. This standard was devised by Eldersveld (1949) and is discussed in Note 8.

at Glasgow University Library on October 15, 2012uar.sagepub.comDownloaded from

http://uar.sagepub.com/

Kennedy’s victory precipitated an agonizing debate within the Republican Party over what to do to close the “big-city gap.” Senator Barry Goldwater publicly asserted that the Republican Party should stop trying to attract bloc support from Negro and other minority groups concentrated in cities (Rourke 1965, 157). With Goldwater’s nomination in 1964, the Republicans pretty much wrote off the urban vote. Indeed, since 1956, when Eisenhower won pluralities in 6 out of the 12 cities in our sample, none of our sample cities has voted Republican in a presidential election.

Notwithstanding the importance of urban votes to the Democrats, Table 2 demonstrates that the power of cities to influence presidential elections has declined since the Roosevelt-Truman victories. Our sample cities were important for the victories of Carter in 1976 and Clinton in 1992 and 1996, but the number of electoral college votes “won” by cities has generally declined (see Figure 2).9 In every election from 1936 onward, with the excep- tion of six cities in 1956, positive figures in Table 2 represent instances when cities made a contribution to the margin of victory of a Democratic candidate within the state. All of the zeroes that appear in Table 2 since 1928, except in 1956, indicate instances when cities produced Democratic presidential plu- ralities while their respective states went Republican. In 1972, every one of Eldersveld’s (1949) sample cities voted Democratic, and all of their

78 URBAN AFFAIRS REVIEW / September 1999

TABLE 1: Sample City Proportion of State Actual Electorates

Election Year

1920 1932 1940 1948 1956 1964 1972 1980 1988 1996

New York 44.1 46.8 51.2 50.6 44.7 41.7 34.6 31 31.2 32.1 Chicago 37.5 40.9 44.9 46.5 37.7 34.2 28 23.5 23.4 20.9 Philadelphia 22.6 21.2 21.8 23.5 19.5 18.9 17 15.8 14.9 11.7 Pittsburgh 5.6 5.6 7.5 7.2 6 5.4 4.3 3.8 3.5 2.9 Detroit 26.8 25.7 26.7 31.8 25.5 21.3 14.8 10.4 8.4 8 Cleveland 8.3 9 11.7 10.6 9.1 7.6 5.4 4.3 3.6 2.9 Baltimore 51.4 48.3 47.9 42.3 34.2 28.4 19.5 17.2 13.6 9.6 St. Louis 21.2 22.2 21.9 21.8 18.1 14.7 10.4 8.1 7.2 5.5 Boston 16.7 16.3 16.9 16.2 13.2 10.9 8.6 7.1 7.1 6.8 Milwaukee 17.8 15.7 19 20.3 18.9 17.7 14.1 12.6 12.2 9.5 San Francisco 15.6 9.8 9.5 8.4 6.2 4.6 3.6 3 2.8 2.6 Los Angeles 27.3 42.8 43.3 42.9 18.5 15.3 12.5 10.7 10.4 7.5

Sample city votes as percentage of national electorate 15.8 18.3 20.7 21.3 15.7 13.2 10.2 7.6 7.1 5.9

at Glasgow University Library on October 15, 2012uar.sagepub.comDownloaded from

http://uar.sagepub.com/

respective states went Republican. In 1984, Ronald Reagan nearly duplicated Nixon’s feat, losing every city but winning every state in our sample, except Massachusetts.

REVISING THE CONVENTIONAL WISDOM

When we update Eldersveld’s (1949) data and analyze it, the results at first seem to confirm the conventional explanations of the decline of the political significance of cities. The conventional account emphasizes that the main reason for the declining electoral influence of central cities is the flight to the suburbs. In 1950, according to the U.S. Bureau of the Census, the population of all of Eldersveld’s sample cities accounted for 57.8% of the total popula- tion of their metropolitan areas; by 1990, these central cities accounted for only 27.7% of that figure. It is well understood that suburbs tend to favor Republican presidential candidates, but the opposite is true for cities. Increas- ingly, since 1948, large Democratic pluralities in central cities have not been able to make up for the rising tide of suburban and rural Republican votes, although it seems that this has not been for lack of effort. As central-city elec- torates have shrunk, they have become proportionally more Democratic. As Schneider (1992) noted, however, suburbs are simply growing faster than cities

Sauerzopf, Swanstrom / THE URBAN ELECTORATE 79

Figure 3: Sample City Electorate as a Percentage of the National Electorate, 1920-1996

at Glasgow University Library on October 15, 2012uar.sagepub.comDownloaded from

http://uar.sagepub.com/

TABLE 2: Sample City Percentage of State Party Pluralities, 1920-1996

1920 1924 1928 1932 1936 1940 1944 1948 1952 1956 1960 1964 1968 1972 1976 1980 1984 1988 1992 1996

New York 40.5 15.8 0 145.9 163.3 320.6 244.9 0 0.29 0 206.2 51.8 187 0 248.2 0 0 253.9 88.7 64.4 Chicago 41.5 41.2 4.7 55.5 77.8 326 298.6 926.5 0 R5 5151 75.8 0 0 0 0 0 0 89.7 74.1 Philadelphia 30.5 29.5 14.6 R44.9 31.6 63.3 142.9 0 0 R20.5 285 29.6 160.4 0 207.7 0 0 0 67.8 77.3 Pittsburgh 7.6 2.3 0.7 0 15.4 24.9 54.3 0 0 0 78.6 8.3 41.5 0 33 0 0 0 16.6 14.2 Detroit 31.9 30.3 15.5 65.2 57.7 0 908.7 0 0 0 466.4 38 95.4 0 0 0 0 0 98.5 51.4 Cleveland 12.7 0 0 56.7 26.3 104.8 0 1228 0 0 0 19.3 0 0 901.8 0 0 0 101.1 28.5 Baltimore 70.9 71.4 11.7 62.3 71.1 75.5 231.8 0 0 R20.1 115.5 47.7 483.9 0 111.4 294.2 0 0 53.7 38.8 St. Louis 37.5 58.9 0 22.4 32 74.7 152.2 38.2 0 1811 1011 29 0 0 85 0 0 0 31.4 50.3 Boston 7.9 4.3 582.4 161.9 65.5 65.7 71.9 58.2 0 0 28.3 15 18.5 0 12.1 465.2 79.1 29.1 14.1 10.8 Milwaukee 7 M20.2 0 27 32.7 280.8 0 96.4 0 R4.3 0 28.9 0 0 201.5 0 0 120.2 77.3 38.7 San Francisco 15.9 1.6 0 15.5 14.1 12 15.8 55.6 3 R2 0 10.7 0 0 0 0 0 0 11.9 11.5 Los Angeles 30.8 58.8 55.5 37.8 43 47.7 46.2 44.4 5.6 R2.7 0 24.2 0 0 0 0 0 0 28 25.9 Mean city % 27.8 31.4 17.2 64.1 52.5 126.9 216.7 347.7 3 266.6 917.9 31.5 164.6 0 225.1 379.7 79.1 134.4 56.6 40.5 Number of

cities 12 10 6 10 12 11 10 7 3 7 8 12 6 0 8 2 1 3 12 12

NOTE: This table shows sample city presidential party pluralities as percentages of sample state pluralities. Zeros indicate instances when state pluralities and city pluralities were for different parties. For 1920 to 1928, all figures indicate sample city support for Republican pluralities, except where indicated by an M for LaFollette Progressive and D for Democratic pluralities. For 1932 to 1996, all nonzeros indicate sample city support for statewide Democratic pluralities, except where indicated by an R for Republican pluralities. All zeroes for this period indicate instances when states produced Republican pluralities despite sample city Democratic pluralities.

at G lasgow

U niversity Library on O

ctober 15, 2012 uar.sagepub.com

D ow

nloaded from

http://uar.sagepub.com/

are growing more Democratic. The vaunted “suburban strategy” that Clinton followed in his recent victories is rooted in basic demographics.

A closer examination of the data, however, suggests that the conventional wisdom is too simplistic. Although the population of our 12 cities has declined, the total Democratic presidential votes coming out of them since 1968 has remained remarkably stable (see Figure 4). There are two main rea- sons for this. First, 4 of our cities located on the two coasts (New York, Bos- ton, Los Angeles, and San Francisco) reversed the postwar downward trend and gained population in the 1980s, largely because of immigration. In 1990, for example, 28% of New York City’s population was foreign born. The rate at which immigrants naturalize, of course, varies, and studies show that immigrants who become citizens tend to vote at low rates (Desipio 1996). In the long run, however, just as immigration fueled the political rise of cities at the beginning of the twentieth century, it is doing the same thing at the end of the century (New York City’s proportion of state actual electorate actually increased between 1980 and 1996; see Table 1).

A second reason why Democratic votes have remained stable in cities is because the urban electorate has voted progressively more Democratic even

Sauerzopf, Swanstrom / THE URBAN ELECTORATE 81

Figure 4: Total Democratic Presidential Votes for Sample Cities and States, 1956-1996 NOTE: Vote tallies are in millions of votes cast for president.

at Glasgow University Library on October 15, 2012uar.sagepub.comDownloaded from

http://uar.sagepub.com/

as the rest of the nation has tended to vote more Republican in presidential elections. A striking trend that emerges from our data is the increasing diver- gence between the voting behavior of cities and the rest of the nation. We can see this clearly here when we examine the index of difference (see Figure 5). The index of difference measures the magnitude of the difference between the partisan division of the vote in cities and in the rest of the state. The higher the final score in a given year, the greater the difference in partisan voting behavior between the city and noncity vote. For instance, if there were no dif- ference between the partisan presidential voting patterns of city and noncity voters, the index would be zero. If, on the other hand, all city voters voted Democratic and voters in the remainder of the state voted Republican, then the index of difference would be 200.

Figure 5 shows that since 1952, the index of difference has increased markedly. In comparatively good years for Republicans, the index of differ- ence is quite high. This illustrates dramatically that Republican presidential victories have occurred, since 1968, in utter opposition to the preferences of the urban electorate.

One would expect the index of difference to fall in years when the Demo- crats do comparatively well, winning more votes outside the cities. However, even in elections when the Democrats win most or all of the states covered by our data, the index generally remains high, reaching its highest point in the 1996 election. Urban voters are increasingly “eccentric,” responding differ- ently to political cues than voters in the rest of the nation. “Solid cities” have largely replaced the “solid South” as the bedrock of the Democratic coalition. Contrary to conventional wisdom, which emphasizes that elections are won or lost in the suburbs, central-city votes are still crucial to Democratic victo- ries in presidential elections (see Figure 2).

According to our analysis, our sample of 12 cities was controlling (con- tributing more than 50% of the statewide plurality) in states representing 138 electoral votes in 1992 and 107 votes in 1996. In 1992, for example, New York City provided 89% of Clinton’s nearly 1 million vote statewide margin. Indeed, Clinton lost (narrowly) to Bush in suburban Long Island (Nassau and Suffolk counties). Obviously, if we went beyond our limited sample to look at, say, all central cities with populations more than 100,000, cities would have been decisive in delivering many more electoral votes into the Demo- cratic column. Although it would be exaggerating our point, it would be pos- sible to turn the conventional wisdom on its head (at least for the 1992 and 1996 elections): How could the Republicans have expected to win the presi- dency when they were incapable of seriously contesting for votes in the big cities?

82 URBAN AFFAIRS REVIEW / September 1999

at Glasgow University Library on October 15, 2012uar.sagepub.comDownloaded from

http://uar.sagepub.com/

Given the fact that central-city voters increasingly vote differently from the rest of the nation and increasingly find themselves on the losing side, per- haps it is not surprising that the propensity for people in cities to vote has declined relative to votes in the rest of the state (see Figure 6). The declining electoral influence of cities is not just due to suburbanization but to demobili- zation of the urban electorate. The effect is significant. Our data show that in 1996, although the total population of voting-age persons in our sample cities represented a healthy 17.4% of the sample state figure, our city voters accounted for only 13.5% of the sample state total.

Sauerzopf, Swanstrom / THE URBAN ELECTORATE 83

Figure 5: Index of Difference, 1956-1996 NOTE: The index of difference measures the gap between the partisan division of the sample cit- ies and the party split in the rest of the sample states in presidential elections. The index is calcu- lated by pooling Democratic and Republican votes for president for all of the sample cities and for all of the sample states minus sample city returns. The percentage of the pooled sample city presidential vote that went Democratic is subtracted from that which went Republican. The same is done for the pooled presidential votes for sample states minus the sample city returns. The two figures are then either added together if the sample city and state presidential votes were won by different parties or subtracted one from the other if the same party won both the city and the state. Index of difference = {(Extracity state % R – Extracity state % D) + or – (City % R – City % D)}. The index of difference was developed by David Olson in 1992, when he was a doctoral student in the Political Science Department at the State University of New York at Albany.

at Glasgow University Library on October 15, 2012uar.sagepub.comDownloaded from

http://uar.sagepub.com/

For the most part, Democrats have been able to take the urban vote for granted because there is nowhere else for it to go. Although our data do not enable us to determine the causes of central-city turnout decline, they do sug- gest that the Democrats should not take their urban base completely for granted. Clinton’s “suburban strategy” may have political costs. As Demo- cratic candidates have distanced themselves from urban policies and sym- bols, the turnout gap between cities and suburbs has grown. Even as central-

84 URBAN AFFAIRS REVIEW / September 1999

Figure 6: Sample City and State Relative Propensity to Vote, 1952-1996 NOTE: The relative propensity to vote compares sample city and state voter turnout rates ad- justed for election-to-election changes in national turnouts. Here, in any given election, the na- tional turnout is set at 1, and sample turnouts are recalculated proportionally. Sample state and city turnouts compare pooled votes for president to pooled voting-age populations for all sample cities and states. Adjusted sample state turnout rates do not include sample city population and voting figures. Nardulli, Dalager, and Greco (1996) used the relative propensity to vote in their comparative study of turnouts. Voting-age population (VAP) figures are calculated by straight- line extrapolation between VAP figures from decennial Census of Population and Housing re- ports. VAP figures through 1968 are based on a 21-year-old voting age.

at Glasgow University Library on October 15, 2012uar.sagepub.comDownloaded from

http://uar.sagepub.com/

city pluralities contributed significantly to both of President Clinton’s victo- ries (see Figure 2), city turnout rates declined sharply in 1992 and 1996 rela- tive to their respective states and the nation as a whole (see Figure 6). These trends, taken together, suggest diminishing returns for an electoral strategy that does not engage central-city voters.

In a more catastrophic scenario for the Democrats, the alienation of the urban electorate could be mobilized by a charismatic third-party candidate. According to aNew York Times/CBS voter survey, in 1992 Clinton actually lost the white vote 39% to 41%. Nationwide, Clinton won 82% of the black, largely urban, vote. If Jesse Jackson had run as an independent in 1992, George Bush probably would have been elected to a second term.10

In short, our data suggest that cities are both more important to the Demo- crats and more volatile than the conventional wisdom suggests. There may be costs to the Democrats’suburban strategy: declining turnout among potential Democratic voters in the cities.

A CALL TO EXAMINE THE SUBURBAN VOTE AND CONTEXTUAL EFFECTS

Although central cities no longer have the power to determine most presi- dential elections, they are still essential to Democratic presidential victories. If anything, our research on voting in central cities underestimates their importance. Our dichotomy between the central-city electorate and the rest of the state, however, fails when examining the most rapidly growing seg- ment of the statewide vote: the metropolitan voters outside central cities. At the same time that cities have declined in relative importance, metropolitan areas have grown. Nardulli, Dalager, and Greco (1996) found that in 1990, the nation’s 32 largest metropolitan areas constituted more than 40% of the national electorate. Therefore, investigating ways in which metropolitan areas structure elections is critical to understanding national electoral politics.

The first issue that should be examined is where to draw the line between the “urban” and “suburban” voters. The conventional wisdom naturalizes this division, assuming that all suburban voters share something essential in com- mon and are fundamentally distinct in political orientation from urban voters. Reed and Bond (1991, 735) referred to this as the “mystified notion of the suburbs as a new, coherent political constituency.” In fact, “cities” and “sub- urbs” are problematic social constructions. The current dividing line between cities and suburbs is arbitrary—being based, as it is, on the jurisdictional boundaries between the original central city and everything else in the metro- politan area. Although some cities are elastic, expanding their boundaries as

Sauerzopf, Swanstrom / THE URBAN ELECTORATE 85

at Glasgow University Library on October 15, 2012uar.sagepub.comDownloaded from

http://uar.sagepub.com/

their metropolitan area grows, others, including most of the cities in our sam- ple, are inelastic, hemmed in by independent suburban municipalities (Rusk 1993). In these cases especially, as metropolitan areas continue to become more urbanized, characteristics associated with central cities spill over into neighboring suburbs.

Although cities increasingly resemble the conventional view, with minor- ity populations, social problems, and physical decline concentrated within their borders, suburbs are becoming more varied. When estimating the role that metropolitan place plays in electoral politics, one too often fails to con- sider the fact that many suburbs do not resemble the conventional image of leafy, white, middle-class, residential refuges. As the dichotomous assump- tions of the conventional view are challenged, so are the consequences of this perspective for cities, metropolitan areas, and the Democratic Party.

Weiher (1991) explored the fantastic proliferation of incorporated municipalities surrounding most of the nation’s large central cities. As he showed, these suburbs over time have developed increasingly distinct eco- nomic and social characteristics. These characteristics structure the cognitive map, Weiher argued, that prospective residents use to inform their moving decisions. Individual residential choices accumulate to further cement the identities of specific suburbs. The result is a sociopolitical fracturing of met- ropolitan areas. This fragmentation, as opposed to stratification, is character- ized by substantial variation among suburbs. However, as variation among suburbs is increasing, many suburban municipalities are becoming more homogeneous internally. The result of this fragmentation may have serious political consequences as identification with municipalities supersedes understandings of metropolitan areas as interdependent polities.

The suburbs that represent the most critical problem for the conventional view are those that more closely resemble central cities, typically inner-ring suburbs. Many inner-ring suburbs—such as Harvey, Illinois; Lackawanna, New York; Compton, California; and East Cleveland, Ohio—have more in common with their central cities than they do with the stereotypical middle- class suburbs. Orfield (1997) has hypothesized that the “swing” votes in recent presidential elections have not been in the suburbs generally but in the inner-ring, distressed (largely white, working-class) suburbs.11

Our analysis provides support for those scholars who argue that the cur- rent period is not characterized so much by a Republican realignment as by dealignment, the inability of either party to assemble a permanent winning coalition.12Under dealignment, the ability of parties to mobilize voters weak- ens, and voters withdraw from politics as such. Our results, which show declining voter turnout both by urban and suburban electorates, support those who argue that the current period is characterized more by dealignment than

86 URBAN AFFAIRS REVIEW / September 1999

at Glasgow University Library on October 15, 2012uar.sagepub.comDownloaded from

http://uar.sagepub.com/

realignment—at least in metropolitan areas outside the South. Future voting research needs to disaggregate the suburban vote. Inner-ring voters may be cross-pressured between economic concerns that pull them toward the Democratic Party and social concerns, especially about race and crime, that draw them toward the Republicans. By breaking out the inner-ring subur- ban vote, researchers could test the hypothesis that this is where voter vola- tility (switching parties) and withdrawal (declining turnout) are centered— differentiating them from central cities and outer-ring suburbs.13

The second major thrust of electoral research on urban areas should be to determine whether there is an independent “contextual effect” of place on voting behavior. The aggregate data used here only allow us to describe vot- ing patterns by place but not to determine why people vote or do not vote the way they do. The results we have reported could be an artifact of the individual-level characteristics of central-city and suburban electorates (e.g., their class or racial background). On the other hand, the results could also be a result of the social networks, sources of information, and reference groups rooted in places. Scholars have hypothesized, for example, that the move from rural areas to cities caused people to become more conscious of the interdependencies of society and the need for government to protect people from the destabilizing and damaging effects of industrial capitalism (Lubell 1965, 47). Similarly, others have argued that the suburban context has a con- servativizing effect, pulling erstwhile Democratic voters toward the Republi- cans: As white ethnics migrated to the suburbs, they became home owners and lived more privatized lifestyles, making them more susceptible to Repub- lican appeals that favor the market over government (Harris 1954; Greenstein and Wolfinger 1958; Schneider 1991, 1992).

More research needs to be done on the contextual effects of different met- ropolitan residential environments. Aggregate data can be used to test for contextual effects by controlling for income, education, race, and other individual-level variables and seeing if place still has independent effects on voting behavior.14 The best data, however, for examining contextual effects are individual-level data (e.g., from sample surveys in which each case has been geocoded so that researchers can move back and forth between individ- ual and contextual levels of analysis) (Huckfeldt and Sprague 1993).

In one of the only studies of contextual effects of central cities, Wolman and Marckini (1997) found that after controlling for a wide range of other characteristics, central-city congressional representatives voted differently than non-central-city representatives. Apparently, the central-city context makes representatives vote more liberal. In one of the few studies of the effect of suburban contexts on voting behavior, Oliver (1994, 1996) has used geo- coded individual-level data to show that political participation declines,

Sauerzopf, Swanstrom / THE URBAN ELECTORATE 87

at Glasgow University Library on October 15, 2012uar.sagepub.comDownloaded from

http://uar.sagepub.com/

controlling for individual-level variables, in homogeneous suburban con- texts. Oliver’s research suggests, beyond selection and rational interest struc- turing, that residence within a specific kind of place has a transformative effect. For Oliver, not only do some middle-class residents select suburbs as a form of escape from dynamic urban political environments (as Schneider 1991, 1992 argued) but residents are also depoliticized by the socioeconomic homogeneity and lack of significant political conflicts associated with these places and their reform-style municipal governments.

If one takes seriously the potential of metropolitan places to exert politi- cally transformative effects on their residents and extend it to other metro- politan municipal types, then understanding how metropolitan areas struc- ture political attitudes and behaviors becomes a critical project, central to understanding national political divisions.

NOTES

1. The seminal statements on realignment theory come from Key (1955) and Burnham (1970).

2. Realignment theory applies to congressional as well as presidential elections. In this article, we deal only with presidential elections. For recent analyses of the decline of urban rep- resentation in the U.S. Congress, see Caraley (1992, 19-23), Wolman and Marckini (1997), and Cook (1997). It is important to note, however, that big cities were consistently underrepresented in Congress and state legislatures until the Supreme Court reapportionment cases of 1962-1964. Therefore, until this time, cities viewed presidential elections as the best way to make their con- cerns heard in national politics (Barone 1990, 106).

3. Clinton knows William Julius Wilson and is on record endorsing Wilson’s views on the problem of concentrated urban poverty (“A Visit with Bill Clinton” 1992). Likewise, Wilson, in a 1992New York TimesOp Ed piece, praised Clinton’s broad-based policy approach (Wilson 1992). Wilson, however, later criticized Clinton’s signing of the welfare reform act (Personal Responsibility and Work Opportunity Act of 1996).

4. Two notable exceptions are Lubell (1965), who reported separate city returns for 12 cen- tral cities from 1920 to 1964, and Nardulli, Dalager, and Greco (1996), who reported results for 32 center cities from 1828 to 1992. The weakness of the latter data set is that for 10 cities, county data are used to represent the city vote.

5. The main published source of presidential election returns is the volumes ofCongres- sional QuarterlyElection Research Center’sAmerica Votes(Scammon, McGillivray, and Cook 1956-1994). These report results for states and counties but not cities separate from counties. The collection and storage of election data in the United States are remarkably backward. Lack- ing a national repository of historical election results, the data are controlled by technologically backward and often patronage-ridden county boards of election.

6. The data used for all tables and graphs for the period from 1920 to 1948 are from Elders- veld (1949). Returns from 1952 to 1992 are from a variety of sources. All state and national elec- tions returns are from Scammon, McGillivray, and Cook (1994). City election data for Milwau- kee, Cleveland, Los Angeles, and Pittsburgh for this period are from county legislative manuals

88 URBAN AFFAIRS REVIEW / September 1999

at Glasgow University Library on October 15, 2012uar.sagepub.comDownloaded from

http://uar.sagepub.com/

and other boards of election records. Presidential returns for all other cities for this period are from America Votes(Scammon, McGillivray, and Cook 1956-1994). Calculations that require total sample city presidential votes are based on total votes cast for president. In a few cases, however, total votes cast for major party presidential candidates had to be used as an approxima- tion because returns for minor candidates were not available. All state and national votes cast for president in 1996 are from the Federal Election Commission. Presidential returns for all sample cities for 1996 were originally obtained from newspaper accounts. We want to thank the follow- ing individuals who helped us to collect these data: Larry Bennett, Barbara Ferman, Robin Jones, Robin Boyle, Dennis Keating, Patrica Atkins, Meredith Ramsay, Marc Levine, Rich DeLeon, Ali Modarres, and Kenny Thomas. We were later able to confirm and edit these 1996 returns based on returns as provided by local and state election offices for all of the cities except Phila- delphia, Baltimore, St. Louis, and San Francisco. Presidential returns for all sample cities are from the central cities themselves, with one exception: Eldersveld (1949) could not obtain elec- tion returns for the city of Los Angeles and so substituted Los Angeles County returns. However, Los Angeles election returns from 1952 to 1996 are for the city of Los Angeles.

7. Alvarez and True (1973) argued that urban ethnics were aligned toward the Democrats throughout the period from 1896 to 1924.

8. Eldersveld’s (1949) criterion was based on the idea that if the city vote had swung in the opposite direction, then the state would have been won by the other party. The idea that city votes could swing from one party to the other is questionable, but we think Eldersveld’s criterion is nevertheless a good general indicator of the electoral significance of cities.

9. Regional population shifts away from the older industrial states are a small part of the reason for the declining influence of our sample cities. The 10 states representing our 12 cities reached a peak of 233 electoral votes in 1960, declining to 215 electoral votes following the reap- portionment based on the 1990 census. The only state in our sample to gain electoral votes during this period was California.

10. Recently released tapes of Richard Nixon record him in 1972 discussing a scheme to finance an independent black candidate. Nixon aide H. R. Haldeman discussed underwriting Shirley Chisolm in the Democratic primaries but then recommended another course. “The argu- ment is that if we can launch a Jesse Jackson or somebody outside the party we’re better off, we can keep them bought.” Nixon replied, “Yeah.” Haldeman then proposed paying Jackson $10,000 for every percentage of the vote he gets (quoted in Feinsilber 1996).

11. There is a growing appreciation of the distinctiveness of these places. See McCormick and McKillop (1989) and Minerbrook (1992).

12. To sample the debate on realignment versus dealignement, see Niemi and Weisberg (1993).

13. Once again, the difficulty in disaggregating the suburban vote is the lack of published voting statistics by municipality. One notable exception is the Record on American Democracy (ROAD) data set put together by King et al. (1997) at Harvard University. For the period from 1984 to 1990, they collected election data for all federal and statewide elections broken down to the precinct level (approximately 170,000 nationwide). The data also were aggregated by “major civil division,” which for most states allows analysis of voting results by suburban municipality. The data set includes 1990 census data. The data also include geographic boundary files so that maps can be drawn. The entire data set is available free on the World Wide Web at http://data.fas.harvard.edu/ROAD/.

14. Because of the small size of ourN(12 cities), using regression analysis to test for contex- tual effects in our case would have been highly problematic.

Sauerzopf, Swanstrom / THE URBAN ELECTORATE 89

at Glasgow University Library on October 15, 2012uar.sagepub.comDownloaded from

http://uar.sagepub.com/

REFERENCES

Alvarez, D. J., and E. J. True. 1973. Critical elections and partisan realignment: An urban test case.Polity 5:563-76.

Andersen, K. 1979.The creation of a democratic majority 1928-1936. Chicago: Univ. of Chi- cago Press.

Barone, M. 1990.Our country: The shaping of America from Roosevelt to Reagan. New York: Free Press.

Burnham, W. D. 1970.Critical elections and the mainstreams of American politics. New York: Norton.

Caraley, D. 1992. Washington abandons the cities.Political Science Quarterly107 (1): 1-30. Cook, R. 1997. Rare combination of forces may make history of ‘94.Congressional Quarterly

Weekly Report53:1076-81. Degler, C. N. 1964. American political parties and the rise of the city: An interpretation.Journal

of American History51(1): 41-59. Desipio, L. 1996. Making citizens or good citizens? Naturalization as a predictor of organiza-

tional and electoral behavior among Latino immigrants.Hispanic Journal of Behavioral Sci- ences18:194-213.

Edsall, T., and M. D. Edsall. 1991.Chain reaction: The impact of race, rights, and taxes on American politics. New York: Norton.

Eldersveld, S. J. 1949. The influence of metropolitan party pluralities in presidential elections since 1921: A study of twelve key cities.American Political Science Review43 (6): 1189-1206.

Feinsilber, M. 1996. Nixon tapes reveal talk of financing a black candidate in ‘72.Albany Times Union, 28 November.

Greenberg, S. B. 1995.Middle class dreams: The politics and power of the new American major- ity. New York: Times Books.

Greenstein, F., and R. Wolfinger. 1958. The suburbs and shifting party loyalties.Public Opinion Quarterly22:473-82.

Harris, L. 1954.Is there a Republican majority?New York: Harper & Row. Huckfeldt, R., and J. Sprague. 1993. Citizens, contexts, and politics. InState of the discipline II,

edited by Ada Finifter, 281-303. Washington, DC: American Political Science Association. Key, V. O., Jr. 1955. A theory of critical elections.Journal of Politics17:3-18. King, G., B. Palmquist, G. Adams, M. Altman, K. Benoit, C. Gay, J. B. Lewis, R. Mayer, and

E. Reinhardt. 1997.The record of American democracy, 1984-1990. Produced by Harvard University. Ann Arbor, MI: Interuniversity Consortium for Political and Social Research [distributor].

Lubell, S. 1965.The future of American politics. New York: Harper & Row. MacRae, D., and J. Meldrum. 1960. Critical elections in Illinois: 1888-1958.American Political

Science Review54:669-83. McCormick, J., and P. McKillop. 1989. The other suburbia: An ugly secret in America’s suburbs:

Poverty.Newsweek, 26 June, 22-24. Minerbrook, S. 1992. A tale of two suburbias: The decline of blue-collar suburbs and growth of

“edge cities” create a new kind of isolation.U.S. News and World Report, 9 November, 32-40.

Mollenkopf, J. H. 1983.The contested city. Princeton, NJ: Princeton Univ. Press. Nardulli, P. F., J. K. Dalager, and D. E. Greco. 1996. Voter turnout in U.S. presidential elections:

An historical view and some speculation.PS: Political Science and Politics29:480-90.

90 URBAN AFFAIRS REVIEW / September 1999

at Glasgow University Library on October 15, 2012uar.sagepub.comDownloaded from

http://uar.sagepub.com/

Niemi, R. G., and H. F. Weisberg. 1993. Dealignment and realignment in the current period. In Controversies in voting behavior, 3d ed., edited by R. G. Neimi and H. F. Weisberg, 321-32. Washington, DC: Congressional Quarterly.

Oliver, J. E. 1994. City size and political participation in the fragmented American metropolis. Paper presented at the annual meeting of the American Political Science Association, New York, September.

. 1996. The influence of social context on patterns of political mobilization. Paper pre- sented at the annual meeting of the American Political Science Association, San Francisco, September.

Orfield, M. 1997.Metropolitics: A regional agenda for community and stability. Washington, DC: Brookings Institution.

Reed, A., Jr. and J. Bond. 1991. Equality: Why we can’t wait.Nation, 9 December, 733-37. Rourke, F. E. 1965. Urbanism and the national party organizations.Western Political Quarterly

18:149-63. Rusk, D. 1993.Cities without suburbs. Washington, DC: Woodrow Wilson Center Press. Scammon, R. M., A. McGillivray, and R. Cook. 1956-1994.America Votes. Washington, DC:

Elections Research Center, Congressional Quarterly. Schneider, W. 1991. Rule suburbia.National Journal, 28 September. . 1992. The suburban century begins.Atlantic Monthly, July, 33-44. Sundquist, J. L. 1973.Dynamics of the party system: Alignment and realignment of political par-

ties in the United States.Washington, DC: Brookings Institution. A visit with Bill Clinton. 1992.Atlantic Monthly, October. Wilson, W. J. 1987.The truly disadvantaged: The inner city, the underclass, and public policy.

Chicago: Univ. of Chicago Press. . 1992. The right message.New York Times, 17 March. Weiher, G. 1991.The fractured metropolis: Political fragmentation and metropolitan segrega-

tion. Albany: SUNY. Wolman, H., and L. Marckini. 1997. Changes in central city representation and influence in Con-

gress. Paper presented at the annual meeting of the Urban Affairs Association, Toronto, Can- ada, April.

Richard Sauerzopf is a graduate student at the State University of New York at Albany. He is writing his Ph.D. dissertation on voting behavior in inner-ring (white, working- class) suburbs. A case study of the Detroit metropolitan area, the research includes an analysis of aggregate voting, as well as survey data. He is a visiting instructor in the Geography and Urban Planning Department, College of Urban Labor and Metropolitan Affairs at Wayne State University.

Todd Swanstrom is a professor of political science at State University of New York–Al- bany. The coauthor of a critical theme text on American politics,The Democratic Debate (2d ed., Houghton Mifflin, 1998), he has written extensively on urban politics and politi- cal economy. Currently, he is researching the political causes and consequences of class segregation across space in U.S. metropolitan areas and the “nonprofitization” of fed- eral housing policy.

Sauerzopf, Swanstrom / THE URBAN ELECTORATE 91

at Glasgow University Library on October 15, 2012uar.sagepub.comDownloaded from

http://uar.sagepub.com/