The Post-World War II Era

In the years following World War II, there were reasons to presume that conflicts over economic issues were less relevant than they had been in previous decades. While there had been enormous inequality and little income growth during the 1930s, World War II spurred the American economy, and in the 1950s and 1960s, incomes increased steadily and inequality declined as a result. In addition, in the decades between the World Wars, it appeared that Republicans generally had accepted the New Deal and its safety nets, and fundamental conflicts over whether government would play a significant role in society were resolved.

These economic trends were accompanied by analyses arguing that material concerns had become less salient in people's lives. Bell (1962) argued that the great political issues were settled and ideological conflicts were over. Inglehart (1971; 1977; 1990) led the argument that the United States had entered a post-industrial era in which people were more concerned with the quality of life than they were with material concerns (see also Ladd and Hadley 1978). Race, in particular, was posited as a major source of division between the parties (Carmines and Stimson 1989; Edsall and Edsall 1991, 137–153), displacing class divisions (Huckfeldt and Kohfeld 1989, 6–16) as had already been the case in the South for decades (Key 1949). Other observers saw the post-1960s era as a time when cultural issues (including crime, abortion, homosexuality, pornography, and free speech) came to dominate American politics, presumably because economic disparities had grown less politically significant (Wattenberg 1995).

The consequence was a widespread conclusion that political divisions based on class were no longer particularly relevant. Some measured class using self-identified social class (Alford 1963; Flanigan and Zingale 1994) while others used socioeconomic status (Ladd and Hadley 1978).¹

Regardless of the approach, the conclusion was that the political relevance of class divisions had declined (Inglehart 1971; Ladd and Hadley 1978, 73, 195–200, 233–239), especially in presidential elections (Abramson 1974, 102–105; Abramson, Aldrich, and Rohde 1995, 146, 152–153; Beck and Sorauf 1992, 166; Books and Reynolds 1975, 368; Carmines and Stanley 1992, 221–222; Edsall and Edsall 1991, 104; Flanigan and Zingale 1994, 104–105; Glenn 1972; Keefe 1994, 214; Ladd 1978, 98; 1996, 204–206; Lawrence 1997, 40; Trilling 1976, 95–130). Many scholars interpreted the declining significance of class as one aspect of an unraveling of electoral attachments to the major political parties accompanied by a decline in the relevance of traditional political divisions (Aldrich and Niemi 1996; Carmines and Layman 1997).

A Premature Conclusion?

Although at one time American class divisions may in fact have narrowed, that decline may well have reflected a combination of conditions that subsequently have changed. Further, analyses focusing on self-identified class or socioeconomic status²

neglected the role of income, which may be the most appropriate indicator of class (Stonecash 2000). Income is, after all, what provides resources for families and creates opportunity for children within those families.

Class divisions are likely to be limited when inequality is declining and when parties do not differ much on public policy; divisions are likely to be greater when inequality is growing and when the parties differ on policy alternatives. The United States has experienced considerable variation in both inequality and party differences over time, which should create corresponding differences in class divisions.

In brief, inequality and party differences were substantial in the 1930s and 1940s. From the 1950s through the 1970s, both inequality and party divisions were declining. Since then inequality has increased, as have party differences about public policy (McCarty, Poole, and Rosenthal 2006). These long-term changes should have produced corresponding changes in class divisions (Stonecash 2006).

Figure 1 displays the pattern of inequality in the distribution of income, measured by the Gini index,³

for the last century. From 1913 to the early 1940s the Gini index was relatively high, with the apex coming during the Great Depression. During the 1940s, inequality declined and remained low for roughly three decades. Then in the 1970s it gradually began to creep up and subsequently has increased almost every year. If inequality is a necessary precondition for class division, it should have been significant in the 1930s and then declined. Today inequality is clearly on the upswing; meanwhile, we see major debates in Congress on a regular basis about what programs should be cut and who will be hurt, and what tax cuts should be enacted and who will benefit (Brewer and Stonecash 2006, Chapter 3).

Given these variations, the other important matter is whether the parties differ substantially on policy alternatives and whether voters recognize such differences. Previous research shows that differences between the parties follow a pattern very similar to that of inequality (Stonecash, Brewer, and Mariani 2003). Figure 2 presents party differences using the DW-NOMINATE scores developed by Keith Poole. Differences were relatively great during the first half of the 20th century and then gradually declined from the 1940s through most of the 1970s. Then the differences began to increase, and they are now as great as they were a hundred years ago (Gerring 1998, 125–158, 232–253; Stonecash 2006; Taylor 1996). Voters have recognized these growing differences. As Figure 3 indicates, respondents in national surveys increasingly say they see differences between the two parties and care about who wins the presidential election.

Class Divisions in American Politics

Given the variations in inequality and party differences over time, we should expect to see greater class divisions prior to and after the 1950s–1970s. To assess this, we need individual-level income and voting data covering a long time span. There are two data sources for such an analysis: early (1930s–1940s) Gallup Polls and the American National Election Studies (ANES) Cumulative File, which has been conducted since 1948. Although these datasets differ over time, both were collected by reputable organizations and allow us to make a rough assessment of long-term patterns.

In Gallup polls conducted in 1936 and 1940 respondents were not asked about their income, but interviewers classified people—based on their housing, possession of a telephone, car ownership, and the like—into five categories of general affluence: “average plus,” “average,” “poor plus,” “poor,” and “on relief” (Ladd and Hadley 1978, 53).⁴

The ANES surveys have asked respondents their family income. In the ANES surveys, respondents are presented with several income categories and asked to self-place into a category. Based on these responses, individuals in the 1948–2004 cumulative file may be classified by their relative position (percentile) in that year's income distribution.

The ANES Cumulative File presents income using five income percentile categories to classify respondents. The percentile groupings are 0–16, 17–34, 35–62, 63–95, and 95–100. Those in a particular percentile grouping may make more or less in absolute dollars over time, but our concern is their relative economic situation. With that in mind, respondents in the bottom two categories are coded together as lower income, and those in the top two categories are coded together as the highest-income households. The extent of class division is then measured by determining the percentage of the lowest economic category who voted Democratic⁵

and subtracting from that the percentage of the top economic category who voted Democratic. The difference is expressed as the simple difference in percentage points, as is done with other analyses of class divisions (see also Brewer and Stonecash 2001; Stonecash 2000; Stonecash, Brewer, McGuire, Petersen, and Way 2000; and Stonecash and Mariani 2000).

To assess the long-term relationship between income and voting behavior, Figure 4 presents the Gini index (multiplied by 100 to make it comparable to percentages) and the difference between the top and bottom third in voting for the Democratic presidential candidate since 1932. The pattern for 1932 through 1968 roughly squares with conventional wisdom. During the early years of this time span, inequality was high, and so were class divisions. Throughout the 1950s and the 1960s, inequality declines and class divisions also decline relative to the levels that prevailed in the 1930s and the 1940s. After 1970, the divisions increase to the levels that existed in the 1940s, but they declined somewhat in 2000 and 2004.

The dominant argument about class divisions has focused on the post-World War II era. Figure 5 presents ANES data focusing only on those years and including differences for Senate and House elections. Here the post-World War II pattern becomes clearer: as inequality increased after 1970, class divisions steadily increased in voting for all offices—until the 2004 elections.

The Intervening Role of Short-Term Forces

The decline in class divisions in 2000 and 2004 is somewhat puzzling, with inequality continuing to increase and party polarization persisting and growing. As often happens in politics, short-term factors matter, and they are the likely source of this decline. Much analysis is needed to understand these two elections, but two factors are especially important. In 2000 a significant factor may have been that most voters did not recognize just how conservative George W. Bush would be. In 2004 it is clear that voting choices were tied closely to reactions to the Iraq War. The issue divided the electorate strongly and that division almost overwhelmed other sources of political division. As Table 1 indicates, the primary source of division in presidential and House voting was tied to attitudes toward the Iraq War. Approval and disapproval of the war created splits within income groups of almost 70 percentage points. Among lower-income supporters of the war, support for John Kerry was only 11%. Among higher-income opponents of the war, support for Kerry was 72%. There was still an income gap in House and presidential voting, but it was relatively small in comparison to the last two decades. The same kind of split occurred in 1998 over the impeachment of Bill Clinton, the last intervening issue to reduce class divisions.⁶

The 2006 Elections and the Future

The Iraq War will someday be over and the polarizing President George W. Bush (Jacobson 2006) will be in retirement.After these short-term factors fade from the scene, presuming no other equally powerful short-term factors arise, the more enduring and fundamental divisions revolving around class are likely again to become more salient. This is especially so because inequality and party differences are continuing to increase. However, given the continuing prominence of the Iraq War and the fact that Bush will still be in the White House, it is unlikely that class issues will be central to the 2006 elections. Republicans are also acutely aware that the Iraq War is hurting them politically, and they are likely to be careful about taking actions that would add class divisions to the current problems they face. Democrats know the primary weaknesses of Republicans are the Iraq War and George Bush, so even though they will argue that Bush and Republicans have hurt the working class, their focus will be on the most obvious Republican vulnerabilities.