7 - Female Voters and the “Rum and Religion” Election of 1928

The election of 1928 merits our attention as the first of a series of elections that led to the durable New Deal realignment and as an election characterized by an idiosyncratic surge in popular interest and passion fueled by the issues of “rum and religion” (Silva 1962). Resurgent controversy over Prohibition and the nation’s first Catholic major party nominee stimulated a large increase in turnout in the North (among urban immigrants) and in the South (among native-born Protestants). New voters in the North were overwhelmingly Democratic, but opposition to Prohibition overlapped with religious bigotry to create a formidable Republican challenge for Democratic nominee Al Smith in the West and South (Peel and Donnelly 1931). Republican nominee Herbert Hoover ultimately captured more than 60 percent of the popular vote in the last demonstration of Republican hegemony before the emergence of the Democratic coalition that would dominate much of the twentieth century.

The impact of women in 1928 was the subject of considerable speculation, both at the time and by later scholars. Many expected that the central issues of the campaign – Prohibition and religion – were of particular interest to women and would send unprecedented numbers of women to the polls. Commentators and scholars claimed that urban and immigrant women, particularly Catholics, who had lagged behind their rural and native-born sisters in adapting to their new political right, would be motivated to come out to support Smith, their religious and ethnic compatriot. At the same time, rural and native-born Protestant women, long-standing Prohibition advocates, were believed to have been alarmed by the prospect of a Smith presidency, and thus especially motivated to support Hoover in opposition to the Democratic candidate. Indeed, some contemporary and later accounts specifically implicate Republican support among women in the Hoover landslide (see Lichtman 1979).

In this chapter, we review these expectations and then turn to our unique estimates to evaluate them. Our research suggests that both accounts are, in fact, quite accurate – to an extent. Many women entered the electorate for the first time in 1928. In almost every one of our sample states, female turnout increases by at least 5 points, and in the Northeast, turnout gains among women exceed 10 points in Connecticut and 15 points in Massachusetts – a truly extraordinary increase in mobilization. As previous scholars expected, women in different states responded in distinct ways to the presidential options in 1928. In Northeastern states with large immigrant populations, such as Connecticut and Massachusetts, sharp increases in turnout among women, and in Democratic support among women, contribute to large Democratic gains from 1924 to 1928. Outside of the Northeast, women in traditionally Republican states remained strong supporters of the Republican presidential candidate: In some states adding to Republican margins over 1924, but in others states remaining in the Republican column despite a shift in electoral support toward the Democrats. In the traditionally Democratic South and Border states, women moved sharply to the GOP, consistent with an expectation that Smith was a particularly unattractive candidate to Southern women.

Yet, for the most part, women’s turnout and vote choice patterns did not distinguish them from men in 1928. Indeed, owing to turnout differences, changed electoral patterns owe more to male voters overall than they do to female. In every state in which women experienced turnout gains, men’s turnout increased to the same extent. In every state that shifted Democratic, Democrats netted more – and often far more – new votes from men than from women. In all but one state that became more Republican, men were responsible for more new Republican votes than were women. The election of 1928 was characterized by considerable disruption of previous patterns, but it was not female voters who drove or accounted for most of the changes. Long-standing accounts that attribute electoral volatility during this period to less experienced and poorly socialized women (notably, Converse 1972, 1974) are not supported by our analysis. Rather, we find men to be as, or more, responsible for the dramatic partisan changes underway in 1928.

The Election of 1928

The incumbent Republican Party entered the 1928 election with a number of advantages, particularly widespread prosperity (Peel and Donnelly 1931). Several Republicans pursued the party’s nomination when Coolidge declined to run for reelection. The selection of Commerce Secretary Herbert Hoover, yet another candidate more friendly to business interests than to agriculture, was indicative of the extent to which economic elites had come to dominate the GOP (Hicks 1960). In the Democratic Party, on the other hand, the urban interests that comprised an increasingly large component of the party’s coalition in the 1920s were finally able to wrest control of the party. New York governor Al Smith embodied the emerging face of the Democratic Party (and in some ways, of the United States): urban, immigrant, Catholic, and wet (i.e., anti-Prohibition). Smith was extraordinarily polarizing; both within his own party and without, he was either vilified or idolized.

In general, the parties’ nominees differed only slightly on many basic policy issues, such as economic regulation. The parties did offer a clear choice on the central issues of the campaign: “rum and religion” (Silva 1962). Political historians debate the extent to which religion or Prohibition was the overriding issue of the election (see Burner 1986). Al Smith’s Catholicism and opposition to Prohibition were only aspects of a set of qualities – Irish, urban, immigrant, machine partisan – that aroused great distrust and opposition among many Americans, particularly Southerners (an important Democratic stronghold) and farmers (Fuchs 1971; Peel and Donnelly 1931). Iconic journalist Walter Lippmann remarked in 1927 that opposition to Smith was “inspired by the feeling that the clamorous life of the city should not be acknowledged as the American ideal” (quoted in Peel and Donnelly 1931, 99). Influential newspaper editor William Allen White described Smith’s candidacy as a threat to “the whole Puritan civilization, which has built a sturdy, orderly nation” (quoted in Fuchs 1971, 2596).

In one sense, the dominant political outcome in 1928 – the clear electoral monopoly of the Republican Party – represented a continuation from the previous two elections. In other ways, however, 1928 did augur important national shifts. The combination of anti-Smith sentiment in the South and effective pro-Smith Democratic mobilization in the North resulted in large expansion of the electorate – an increase of nearly 7 million voters nationwide over 1924. As a result, turnout nationwide increased from 44 percent to 52 percent, an increase in the total ballots cast for president from 29 million to nearly 37 million. This remarkable expansion in participation in only four years had the potential to radically change the outcome of the election, but the overall Republican advantage over the Democrats was little changed from 1924 and the Republican plurality nationwide remained higher than 6 million votes for the third consecutive election.

This apparent stability in the Republican vote share masks considerable volatility, however. In retrospect, the election of 1928 contained the seeds of the realignment that would, four years later, catapult Democrats into majority party status for at least the next forty years. Smith made significant inroads into the Republican-dominated North, particularly in counties with large Catholic and immigrant populations (Hicks 1960; see also Harris 1954; MacRae and Meldrum 1960). Capitalizing on a revolt among dry southerners, Hoover realized gains in the traditionally Solid South in 1928 (indeed, Hoover carried more than half of the formerly Confederate states), but that support was fleeting. In contrast, Smith’s mobilization of immigrant, urban, and Catholic voters, solidified by Franklin Roosevelt in the 1930s, would permanently reshape the Democratic Party coalition (Degler 1964; Harris 1954; Hicks 1960). One effect of this movement – away from the Democrats in the South and away from Republicans in the North – was to sharply decrease the regional variation in the 1928 presidential vote compared to that which had dominated the previous electoral era (Schantz 1992). These developments were not sufficient to elect Al Smith in 1928, but they laid the foundation for the realignment under Roosevelt in the 1930s.

Expectations for Female Voters in 1928

Many accounts of the election of 1928 identify women as an important part of the surge in electoral participation. The New York Times went so far as to claim that “This year the President of the United States will probably be chosen by women” (McCormick 1928, quoted in Greenlee 2014, 27). Both parties once again exerted considerable effort to appeal to female voters specifically, with Republicans making particularly strong efforts that the Democrats could only attempt to match. The Republican campaign drew notice from The New York Times as being “unusually segregated on gender lines” (quoted in Harvey 1998, 132). In reaching out to women, Republicans highlighted Hoover’s work on behalf of housewives, his support for Prohibition, and his commitment to peace (Greenlee 2014; Harvey 1998). Gender, ethnicity, and religion intersected in important ways. The New York Times recounted a prominent Republican addressing the Women’s Republican Club in Springfield, saying, “I cannot say very much of Mrs. Smith, but if the contest were between Mrs. Hoover and Mrs. Smith...” The writer noted that “There, [the speaker] adroitly left his hearers to draw their own inferences,” the presumption being that an Irish Catholic women did not fit his listeners’ image of an appropriate First Lady (quoted in Peel and Donnelly 1931, 100). Other Republican writers described Mrs. Hoover as “American through and through.” As Peel and Donnelly (1931, 84) note, “The implication was that Mrs. Smith lacked the essential Nordic background, and was somewhat deficient in those social graces and intellectual attainments which characterized the wife of the Republican nominee.”

Democrats also appealed directly to women through women’s offices and campaign committees, female speakers, and gender-specific appeals. As with the Republicans, the Democratic campaign emphasized women’s traditional roles. Democratic nominee Al Smith hailed mothers as “one of our greatest national assets” (quoted in Harvey 1998, 134), and highlighted his stances on such issues as protective labor laws. Even his controversial stance on Prohibition was reframed to appeal to women, with Democrats imploring women to “Think clearly on the Prohibition issue” (Greenlee 2014; Peel and Donnelly 1931). Weekly radio broadcasts sponsored by the League of Women Voters were believed to increase political interest among women, particularly in remote and rural areas (Andersen 1996).

Contemporary observers expected that the central issues of that campaign – Prohibition and religion – would mobilize women on both sides at levels far exceeding previous elections. In The New York Times, for example, headlines such as “Forecasts Big Vote by Women of State” (October 7, 1928) and “This Year’s Woman Vote to Set a High Record” (October 21, 1928) were common (although to be fair, journalists in search of news seemed to expect that every election would be a breakthrough year for female voters). Many states reported increases in female registration.¹

Analysis immediately following the election ratified these expectations. The New York Times reported that

The women’s vote was undoubtedly an important factor. They took a keener interest in the issues than ever before, and they swelled the registration to unprecedented figures. The great majority of women were intent chiefly on saving prohibition, but they also gave their support to Hoover, the executive who enlisted their cooperation in food conservation during the war.²

Explaining the narrow Republican win in Texas, Peel and Donnelly (1931) emphasize the women’s vote in particular (along with rain in the Smith counties, rebellion in the western sections, and a lukewarm endorsement from state leaders). A Democratic Party leader in Minnesota reported his assessment of the impact of female voters in a January 1929 letter to then-New York Governor Franklin D. Roosevelt:

Our worst opposition in this Midwestern country was the women. Republican farmers by the thousands voted for Governor Smith, but their wives and daughters and the great majority of single women voted against Governor Smith, largely because of his stand on the liquor issue – that is particularly true in Minnesota, Iowa, North and South Dakota. I believe if it were not for the Nineteenth Amendment, giving women the right to vote – in other words, if the election had been left to men – Governor Smith would have been elected. You understand, however, that I am not blaming the women except that they were made panic stricken, partly because of the liquor issue, and party on account of Governor Smith’s religion (emphasis ours).³

Many later scholars accepted a significant mobilization of women as part of the conventional wisdom surrounding the presidential election of 1928 (e.g., Matthews 1992). Party realignment theorist Walter Dean Burnham (1974, 1015) writes, “it is widely agreed that female turnout did not begin to move much closer to male participation rates in the United States until the 1928–1936 realignment sequence.” Fellow realignment scholar James L. Sundquist (1983, 192) agrees: “The 1928 increase [in turnout] was probably due mainly to the motivation of many women who, though enfranchised in 1920, had never voted until the spirited contest of 1928 brought them to the polls.”

Consistent with contemporaries who expected most women would support Hoover, scholars describe native-born and rural women as especially mobilized against Smith and in defense of Prohibition and Protestantism in 1928 (Andersen 1996; Burner 1986). While perhaps mobilized earlier and in large numbers than urban and immigrant women, rural and native-born women were certainly not at their limit for mobilization in the mid-1920s. Defending against the specter of an Al Smith presidency may have brought out many previously inactive women, particularly when a well-organized Republican Party specifically sought their votes.

At the same time, many point out that the strong mobilization of native-born and rural women was likely counterbalanced by a surge in turnout among immigrant and urban women. Such women were believed to have been slow to mobilize (compared to native-born Protestant women) in 1920 and 1924, but finally stimulated by their enthusiasm for Smith in 1928 (Andersen 1979; Burner 1986). According to Jeffries (1979, 21), Democrats realized “spectacular” gains among new immigrants in Connecticut, “many of them, women especially, first-time voters.” Burnham (1974, 1015) claims, “This 1928–1936 influx of new voters was heavily concentrated among working-class and women – particularly ethnic women – voters in large industrial states like Pennsylvania and Illinois.” Lubell (1952, 40) asserts

Smith made women’s suffrage a reality for the urban poor. In better income families, women started voting in 1920 as soon as they were granted the privilege; but among the urban masses the tradition that a woman’s place was in the home still held strong until 1928. That year in Massachusetts (which Smith carried along with Rhode Island) the outpouring of women lifted the number of voters by 40 per cent over 1924. The turnout in Boston was 44 per cent heavier.

Empirical verification of these claims has been limited. Focusing on ethnically homogenous neighborhoods in Boston, Gamm (1986; see Chapter 3) observes no particular surge among Italian women, African American women, or Yankee women in 1928, but does find that mobilization of women explains increasing Democratic support in Jewish and Irish neighborhoods. Italian women and African American women were mobilized only in later elections – 1932 and 1936. Consistent with accounts suggesting the earlier mobilization of native-born Protestants, Gamm finds Yankee women had already been integrated into electoral politics in 1920 and 1924.

Most of the claims about the mobilization of women in 1928 rely on data that are rather thin (a report in The New York Times about increased female registration in Catholic precincts in Chicago, for example; Burner 1986), limited to one city (e.g., Boston in the case of both Huthmacher 1959 and Gamm 1986), or in some cases, are made without reference to established fact (e.g., Lubell 1952). All of the claims we have located regarding 1928 ultimately cite to just a small number of sources, each based on a small amount of data, data of questionable interpretation, or no data at all. As later authors cite the earlier authors, and each other, small observations in earlier work becomes generalized conventional wisdom (e.g., Burnham 1974; Sundquist 1983). Given the challenges of data collection during this era (see Chapters 3 and 4), a few scholars have done the best they can to draw reasonable and cautious inferences from the available data, but – given the state of the evidence on which the conventional wisdom is based – to say that the impact of women on the election of 1928 is a settled question would be a considerable overstatement. Our estimates allow us a closer and broader look at the impact of female voters during the momentous election of 1928 than has been previously possible.

The Mobilization of Women in 1928

Previous commentators are correct that women’s turnout rose considerably in 1928. In our sample, the rate of turnout among women grew to 46 percent, an 8-point gain over 1924, when women’s turnout averaged 38 percent. Yet male turnout also posted an 8-point gain, increasing from 67 percent in 1924 to 75 percent in 1928, contrary to accounts that attributed turnout gains in 1928 largely to women (e.g., Sundquist 1983). As a result, the turnout gender gap in our sample remained steady at about 29 points.

Turnout was high in nearly every state. For the first time, the turnout of women (again, conservatively estimated as the proportion of the age-eligible population) exceeded 50 percent in nearly half of the sample states (see Figure 7.1). Democratic-dominated Oklahoma experienced the largest turnout gender gap in the sample (more than 37 points), as it had in 1920, while competitive Missouri again boasted the smallest gap between male and female turnout rates (24 points).

Figure 7.1. Turnout of women and men, 1928.

The level of male mobilization outside of the South in 1928 was truly remarkable – nearing or exceeding 80 percent in every state in our sample. Indeed, while we note the persistent and significant gender gap in turnout, it is worth emphasizing that women’s level of electoral mobilization suffers only by comparison to the extraordinary mobilization of men during this era. In 1980, the first presidential election in which female turnout exceeds male turnout (US Census 1982), overall turnout was 59 percent, a level of mobilization our estimates suggest women in many of our states were approaching in 1928, and indeed, consistent with levels observed among women in competitive Missouri and Kentucky as early as 1920. Given how widespread electoral mobilization was in this period, it is appropriate to acknowledge that women were less mobilized than men. At the same time, we also note how well mobilized new women voters were by the – admittedly low – standards of the modern era.

Were women responsible for the uptick in turnout in 1928? Figure 7.2 reveals that women’s turnout increased over previous years in all but one state. The increase in turnout varied considerably by region and state, ranging from 16.8 points in Massachusetts (a 60 percent increase!) to negligible changes in Iowa and Kansas. The point estimates indicate women’s turnout grew more than men’s in five of our ten sample states, and fell or grew less then men’s elsewhere, but the credible intervals around those point estimates are so wide that we can only conclude with any confidence that men’s and women’s rates of turnout responded similarly to the “rum and religion” election. Contrary to expectations, women were not uniquely mobilized in 1928. Rather, in terms of percentage point increases in turnout, both men and women were mobilized by the 1928 election and largely in the same way and to the same extent.

Figure 7.2. Change in turnout of women and men, 1924 to 1928.

For both men and women, the increase in turnout in our Northeastern states (Connecticut and Massachusetts) stands out as particularly stunning. As we will see (and as others have documented), Al Smith’s candidacy transformed electoral politics in the Northeast, starting with a dramatic increase in voter mobilization (Burnham 1967). These states boasted large immigrant communities: In both Connecticut and Massachusetts, the percentage of the population that was first- or second-generation immigrant was an astounding 65 percent in 1930 (US Census 1932). Contrary to the expectations of some that women, particularly immigrant women, would be responsible for much of the increase in turnout in those states (e.g., Burnham 1974), our estimates suggest that in these heavily immigrant states, both men and women increased their rate of turnout and to the same degree.

Impact of State Context

With the massive partisan and electoral changes underway in 1928, did the political and legal context continue to weigh more heavily on the turnout decisions of women compared to men?

Partisan Context. In some ways, the impact of partisan context remains similar to what we observed in 1920 and 1924 (see Figure 7.3). Competitive states continue to boast the highest rates of turnout among both men and women compared to one-party states, with one-party Democratic states retaining distinctively low turnout levels. However, unlike 1920 and 1924, the only real turnout differences – for women and for men – are between the one-party Democratic states and the two other partisan categories. The turnout of women in competitive states was indistinguishable from that of women in one-party Republican states, and the same can be said of men in those two categories of states. The difference between competitive and one-party Democratic states remains stark. Pooling states by partisan context, 100 percent of the simulations are consistent with women’s and men’s turnout in competitive states and one-party Republican states exceeding that in one-party Democratic states. Consistent with the estimates from 1920 and 1924, the effect of competition was stronger among women: A 28-point difference for men (a 52 percent increase in competitive states over one-party Democratic) versus a 35-point difference for women (a nearly 300 percent increase over turnout in one-party Democratic states). Nearly 90 percent of the simulations are consistent with the conclusion that the difference in turnout is larger for women than for men. Thus, women’s turnout continues to be more affected by party context, as the peripheral voter hypothesis predicts.

Figure 7.3. Partisan context and the turnout of women and men, 1928.

The disappearing gap between turnout in competitive and one-party Republican states is one of many indicators that the 1928 election disrupted established state-level partisan context in ways that previous elections did not. As we explain in Chapter 4, we use a measure of partisan context that taps partisan state legislative lead for 1914 through 1930 to gauge the grassroots partisan context in which women were politically socialized and first cast ballots. Those state categorizations are fairly accurate descriptions of presidential party competition in the states through the 1924 presidential election; that is, Republican presidential candidates tend to win handily in one-party Republican states, while presidential outcomes were closer in competitive states. Electoral change in 1928 is sufficient to break down those associations. Figure 7.4 reports the average margin of victory for the Republican presidential candidate in each of our categories of states in 1920, 1924, and 1928.

Figure 7.4. Partisan context and the Republican margin of victory, 1920 to 1928.

Clearly, 1928 is a major break from what came before. The change in Republican margin of victory in our one-party Democratic states is stunning, moving from a Republican disadvantage in 1920 and 1924 to the largest Republican advantage of states in our partisan categories in 1928. In 1920 and 1924, the Republican margin of victory in competitive states was relatively small, particularly compared to the massive Republican advantage in one-party Republican states. The Republican advantage in previously competitive states more than doubles in 1928 over past years. At the same time, the Republican advantage in previously one-party Republican states in 1928 falls by more than half its level in 1920 and 1924. The change in the one-party Republican states is particularly dramatic in the Northeast. In our four Midwestern one-party Republican states, the Republican advantage in 1928 is 18.4 points; in our two Northeastern states, it is a mere 1.3 points.

Legal Context. Legal restrictions continue to have a large effect on women’s turnout (Figure 7.5). Women’s turnout was 16.3 points (32 percent) lower in states with multiple restrictions compared to those with a residency requirement only, while the similar difference among men is only 12.2 points (16 percent); more than 95 percent of simulations are consistent with the conclusion that female turnout is more affected by the legal context than is male. The relationship between legal context and turnout is smaller in 1928 than in previous elections, however. In 1920 and 1924 the difference in women’s turnout between states with few and many restrictions was consistently around 25 points, compared to about 16 points in 1928. For men the difference was around 19 points, compared to about 12 points in 1928. Nonetheless, the expectation that women’s turnout would be more hampered by legal restrictions than men’s continues to be borne out in 1928.

Figure 7.5. Legal context and the turnout of women and men, 1928.

The Partisan Mobilization of Women in 1928

The 1928 election represented an important break from the past as previously competitive and one-party Democratic states moved toward the GOP and previous Republican strongholds became competitive or even shifted to the other side. What role did women play in this transformation? Overall, both men and women in our sample continued to favor the GOP. As in past years, women were slightly more Republican in their vote choice than were men. In our ten sample states, the Republican vote share averaged 56 percent for men, and 60 percent for women, a difference of 4 points. This was a slight increase in Republican vote share for both men and women over 1924 – a year characterized by a significant third-party vote – up from 53 percent for men (3-point increase) and 58 percent for women (2-point increase).

The state-level data (summarized in Figure 7.6) reveal that this increase in Republican support occurred over a wide range of states. Both of the traditionally competitive states – Democratic-leaning Kentucky and Republican-leaning Missouri – went Republican, and in both, men and women voted Republican to a similar degree. Amazingly, both of the traditionally one-party Democratic states in our sample (Virginia and Oklahoma) moved into the Republican column in 1928; Republican gains in Oklahoma are particularly strong. There is some indication that women in Southern/Border states may have played a distinctive role in this unusual partisan switch; while in Virginia, men and women voted Republican to a similar degree, in Oklahoma, almost 90 percent of simulations (87 percent) are consistent with greater Republican vote share among women compared to men. As many expected, we have evidence that Southern women were particularly mobilized to vote against Smith and for Hoover in 1928.

Figure 7.6. Republican vote share of women and men, 1928.

At first glance, the one-party Republican Midwest (Minnesota, Iowa, Kansas, and Illinois) shows the most stability, staying firmly on the side of the GOP. As we have seen in 1920 and 1924, women in those states continue to be more mobilized by the locally dominant Republican Party than are men, but the differences appear to be narrowing: Using the criteria outlined in previous chapters, women’s Republican vote share exceeds men’s in over 90 percent of our simulations in only one sample state, Illinois. About 88 percent of simulations are consistent with greater Republican support among female voters in Kansas and Iowa. In 1920, Illinois, Iowa, and Kansas met the 90 percent threshold; in 1924, all four of the one-party Republican states in the Midwest did. Not only does our confidence in a partisan gender difference in these states decline for 1928, but the size of the differences also diminishes significantly between 1924 and 1928.

Finally, traditionally one-party Republican states in the Northeast moved strongly toward the Democrats – Massachusetts actually falls into the Democratic column, and Connecticut nearly does so as well. In those two states, men and women’s Republican vote share is indistinguishable, suggesting a similar response to Smith from female and male voters.

Because of the significant third party showing in 1924, tracing the process of change requires attention to shifts in both Republican and Democratic Party vote share, reported in Figure 7.7. Men and women are moving in similar ways in terms of the size of the change in both Democratic and Republican vote share, and the credible intervals are sufficiently wide that we can only conclude with any confidence that they do so to a similar extent.

Figure 7.7. Change in partisan vote share of women and men, 1924 to 1928.

As we have seen, organizing our states by grassroots party context (one-party or competitive), as we have done in previous chapters, is complicated by the changes underway in 1928. Diminishing variation in Republican support is obvious – while in previous years state-level party context strongly predicted presidential vote choice, in 1928, only roughly 3 to 8 percentage points separate male and female voters across the various partisan contexts (see Figure 7.8). The credible intervals indicate that Republican vote share – for both men and women – is indistinguishable across our competitive, one-party Democratic, and Midwestern one-party Republican states in 1928. Only male and female voters in Northeastern one-party Republican states are distinctive – voting more Democratic than men and women, respectively, in the other states, including states traditionally dominated by the Democratic Party!

Figure 7.8. Partisan context and the Republican vote share of women and men, 1928.

Consistent with estimates from 1920 and 1924, the Republican vote share of women and men are similar in competitive states. And despite a dramatic swing to the Democrats, male and female vote choice in the previously one-party Republican states in the Northeast remains indistinguishable as well. But, also consistent with 1920 and 1924, in one-party Republican states in the Midwest, women continue to be more Republican in their vote choice than men. This finding just misses our 90 percent threshold; 89 percent of the simulations are consistent with this conclusion. Persistent and high levels of Republican support among women appear to dampen the surge among men toward Democrat Al Smith in a number of Midwestern states. Previous observers attribute this outcome in 1928 to female opposition to Smith in those states. However, given the similarity to our results in 1920 and 1924, it is possible, even likely, that other longer term factors – specifically, a consistent pattern of loyalty to and/or mobilization of women by the locally dominant party in the Republican Midwest – contributed to this outcome.

There is some evidence that women also are more likely to cast Republican ballots in the one-party Democratic states (all South/Border) in our sample in 1928; 80 percent of simulations are consistent with that conclusion. Although we can be only moderately confident of the result, this greater Republican support among women in traditionally Democratic states suggests loyalty to locally dominant parties (our consistent finding among women in the Republican Midwest in 1920 and 1924) may not have characterized all elections, all contexts, or all times. Consistent with the expectations of many (see Andersen 1996; Burner 1986) that native-born white Protestant women, especially in the South, were mobilized against Al Smith, women in one-party Democratic states in the South and Border regions may have indeed been more likely to cast ballots for Republican Hoover than were men. Our findings for 1928 thus complicate attempts to generalize about the party loyalty of female voters.

The Contributions of Women to the Election of 1928

Partisan vote share can only tell us so much. We now turn to the change in the number of voters – overall and for each major party – in each sample state. We begin with changes in the number of votes cast, reported in Figure 7.9. Both men and women entered the electorate in large numbers in 1928. In most states, the number of new ballots cast by men and women is quite similar, verified by the patterns in the simulations. Given that so many more women were available for mobilization, the fact that in most states as many or more new male voters entered the active electorate in 1928 suggests a substantial and meaningful turnout surge among men.

Figure 7.9. Change in number of votes cast by women and men, 1924 to 1928.

On the whole, Figure 7.9 confirms that about the same number of new ballots came from men and women – perhaps slightly more new ballots from men in Illinois and perhaps slightly more ballots from women in Massachusetts. This outcome is unique to the 1928 election. Of the five elections we examine, the largest new mobilization of women (new votes cast) occurs, unsurprisingly, in 1920, an election in which approximately 1.3 million more new female than male voters were added to the electorate in our ten sample states; 100 percent of the pooled simulations are consistent with more new female than male voters in 1920. Similarly, 86 percent of the simulations are consistent with the same conclusion in 1924. But, in 1928, only 24 percent of the simulations indicate new female voters outnumbered new male voters, suggesting that new male mobilization may have exceeded new female mobilization. (Later elections resume the pattern of larger new female mobilization, with 80 percent and 98 percent of simulations supporting that conclusion in 1932 and 1936, respectively.) As many expected, the “rum and religion” election of 1928 mobilized many new female voters, but it brought as many or more new male voters to the polls. Although previous work (e.g., Burnham 1974; Sundquist 1983) has suggested that women were uniquely or especially responsible for the surge in turnout in 1928, we find that women accounted for a little less than half of the nearly two million additional (or new) ballots cast in the 1928 presidential election in our sample states.

South/Border. In the South and Border states, opposition to Smith was believed to induce many citizens to defect from Democratic loyalty, particularly women. Our estimates of the change in Republican (Figure 7.10) and Democratic (Figure 7.11) votes cast suggest little or no decline in Democratic votes cast among women in our three Southern/Border states (Virginia, Kentucky, and Oklahoma), but a notable Republican mobilization. Although we cannot say definitively, the increase in Republican votes cast by women appears to have been generated largely from new voters (Figure 7.9), given the lack of change in Democratic votes cast and the relatively small number of female Progressive votes available to be recovered from 1924 (see Figure 7.12).

Figure 7.10. Change in number of Republican votes cast by women and men, 1924 to 1928.

Figure 7.11. Change in number of Democratic votes cast by women and men, 1924 to 1928.

Figure 7.12. Change in number of Progressive votes cast by women and men, 1924 to 1928.

Yet, Republicans gain more new male voters than female voters in all three Southern/Border states, especially in traditionally Democratic-dominated Virginia and Oklahoma. As a result, the considerable net gains Republicans experienced in Virginia and Oklahoma (see Figure 7.13) owe far more to male ballots than to female. More than 95 percent of the simulations support this conclusion in Virginia and about 89 percent in Oklahoma. This difference in net Republican gains in Southern states is driven largely by the greater relative Republican mobilization of new male voters compared to female voters (Figure 7.12); the change in number of new Democratic voters is both small and quite similar for men and women (Figure 7.13). Despite a higher percentage of female voters supporting the Republican candidate (recall Figure 7.9), the sheer number of male voters – far exceeding female voters over all in the region – meant that men accounted for much of the large Republican gains in Southern and Border states.

Figure 7.13. Change in Republican margin of victory for women and men, 1924 to 1928.

Northeast. In the Northeast, enthusiasm for Smith was expected to generate large Democratic gains, especially among undermobilized immigrant women (e.g., Jeffries 1979; Lubell 1952). Our data cannot speak directly to the behavior of immigrant (or native-born white, or any other specific subgroup of) women or men. However, as we have noted, our two Northeastern states boast enormous immigrant populations; by 1930, 65 percent of the population of both Connecticut and Massachusetts is first- or second-generation immigrant (US Census 1932).

Our estimates indicate that in the heavily immigrant Northeast, a huge number of women did indeed enter the electorate as Democrats in 1928: Nearly 240,000 new female voters cast ballots in Massachusetts alone (Figure 7.9). Another 40,000 female votes were available from women who voted Progressive in 1924 (Figure 7.12). Of the approximately 280,000 female votes available due to mobilization and defection (from Progressives) of women in 1928, Democrats captured 225,000 or about 80 percent (Figure 7.11). Republicans, by contrast, gained only about 53,000 new female voters (Figure 7.10). Patterns are similar in Connecticut. We cannot determine definitively how many new Democratic and Republican votes can be attributed to mobilization (new voters) or conversion (previous voters for other parties, especially Progressives). But given the estimated number of new female voters (230,000) and estimated decline in female Progressive voters (40,000), mobilization appears to have played the dominant role in changing partisan patterns among women in this region. The number of presidential ballots cast by women increased dramatically in both Connecticut and Massachusetts in 1928, and the vast majority appear to have been cast for the Democratic candidate, consistent with long-held expectations.

Despite this remarkable mobilization of women in the Northeast, it was still men – participating at much higher levels – who accounted for most of the Democratic gains. Fewer men cast new Republican ballots than did women in Massachusetts and Connecticut, but men contribute more new Democratic ballots in both states, especially Massachusetts. Conversion likely played a larger role for men than it did for women in the Northeast: In both states, more male Progressive Party ballots were cast in 1924 than female, and thus many more men were available for conversion in 1928 (Figure 7.11). In Massachusetts, defecting Progressive Party voters accounted for about 97,000 male ballots and new male voters accounted for about 209,000. Of these 306,000 potential new major party votes, then, some 68 percent came from mobilization (compared to our estimate of about 86 percent for women), and 32 percent were available from previous Progressive Party voters (compared to just 14 percent among women). Democrats captured 287,000 of the votes available from new mobilization and Progressive defection, while Republicans increased only by 19,000 votes cast. As in 1924, then, we find more evidence of conversion among men than among women, but mobilization of men in the Northeast in 1928 is still considerable. The consequence is that net Democratic gains (net Republican losses) are much higher among men than women: More than 90 percent of the simulations support this conclusion in Massachusetts and more than 85 percent in Connecticut.

A closer look at Boston both confirms these general conclusions, and provides further verification of the reliability of our estimates. Consistent with Gamm’s (1986) careful analysis of precinct-level voting records (see Chapter 3), our estimates reveal a remarkable level of mobilization in Boston in 1928. Combining turnout and vote choice, we find that nearly half of all age-eligible males are mobilized to support the Democratic candidate in 1928. The comparable number for women in Boston, largely a consequence of lower turnout, is only about 25 percent. Thus, Democratic shifts in the Northeast in 1928 – even where those shifts were largest – appear to be driven by male, not female, voters, contrary to expectations. Certainly there were more men available for conversion – in Massachusetts, nearly 100,000 male Progressive voters alone, compared to just 40,000 female Progressive voters. Yet far more women than men were available for mobilization: We estimate that over 60 percent of men in Massachusetts turned out to vote in 1924, but fewer than 30 percent of women. Despite that fact, men cast nearly equal numbers of new votes in Massachusetts (and Connecticut) in 1928 (Figure 7.9).

Midwest. Our five Midwestern states tell a more complicated story. The general claim has been that rural Protestant women, hostile to urban, Catholic, and anti-Prohibition Smith, shored up Republican support in the Midwest. Our estimates, however, show that while staying solidly in the Republican column, three of our five Midwestern states – Minnesota, Iowa, and Illinois – experienced a net Republican loss among both men and women in 1928 (Figure 7.13). Republicans gained voters in all five Midwestern states (Figure 7.10) but in those three states, Democrats gained far more (Figure 7.11). In Missouri and Kansas, on the other hand, Democratic gains were small relative to Republican gains, and Republicans experienced net gains (Figure 7.13).

What explains these two different patterns? We emphasize, once again, that our state-level summary estimates do not permit us to observe the electoral behavior of any subgroup, and we remain highly cognizant of the ecological fallacy. That being said, we note that our two groups of Midwestern states with distinctive partisan patterns in 1928 are distinguished by the size of their immigrant communities. More than one in three – and often far more – people of voting age in the three Midwestern states that swung toward the Democrats in 1928 were first- or second-generation immigrants: 40 percent in Iowa, 52 percent in Illinois, and 71 percent in Minnesota. The states that experienced a net Republican gain have comparably smaller immigrant populations – about a quarter of the voting-age population in Missouri and Kansas (US Census 1932).

As in the heavily immigrant, traditionally Republican Northeast, in the relatively heavily immigrant, traditionally Republican states in the Midwest (Minnesota, Iowa, and Illinois), women swung not to the Republicans in opposition to Smith, but to the Democrats (Figure 7.13). Like Connecticut, but unlike Massachusetts, all three states remained in the Republican column. As in the Northeast, both parties experience gains among women, but Democrats more so than Republicans. Progressives fared well in these three states in 1924, and so there is more evidence of conversion among women in these Midwestern states than in the Northeast. In Minnesota, for example, there were approximately 100,000 female Progressive voters in 1924, and about 75,000 new female votes cast in 1928. Democrats gain over 110,000 female votes in 1928, and Republicans gain about 65,000. Unfortunately, we cannot determine from our data how many Progressives votes went to Democrats or Republicans (or stayed home), or how the newly mobilized voters were distributed in 1928. We can only conclude that it is likely that both conversion and (to a slightly lesser extent) mobilization produced those new Democratic and Republican female ballots.

Yet, as in the Northeast, as much as women contributed to electoral change in the Midwest, men contributed more. In Minnesota and Illinois, more than 90 percent of the simulations indicate that net Democrat gains (net Republican losses) were larger among men than women. The Democratic Party leader quoted earlier in the chapter was correct to conclude that women were voting somewhat differently than men; but it was only the case that women remained as a relatively stable source of votes for Hoover and were lukewarm supporters of Smith, not that women shifted overwhelmingly in favor of Hoover and that Smith failed to mobilize women. As in the Northeast – but to an even greater extent – the availability of larger numbers of male Progressive voters appears to drive the difference. As we saw in Chapter 6, there were far more male than female Progressive voters in these three states. Those numerous male Progressive voters were available to be converted to major party votes in 1928: nearly 280,000 male Progressive voters in Illinois (compared to about 140,000 among women in that state) and 230,000 in Minnesota (compared to about 100,000 women).

Thus in these three states (Minnesota, Iowa, and Illinois), we observe equal numbers of new male and female voters (Figure 7.9) but many more male 1924 Progressive voters (Figure 7.12). As a result, there were more male voters available to cast ballots for the two major parties (either new voters or former Progressive voters). The number of new Republican votes cast by men and women in these three states are virtually equal. Democrats, on the other hand, gain far more new votes from men than women. In Illinois, women cast approximately 265,000 new Democratic votes in 1928 compared to approximately 470,000 new male Democratic votes. In Minnesota, women cast a little more than 114,000 new Democratic votes, compared to more than 226,000 new Democratic votes cast by men. The result is large net Democratic gains among men, far exceeding the net Democrat gains among women (Figure 7.13).

We cannot determine with certainty how the votes available from men and women – due to new mobilization or Progressive voters without a Progressive option in 1928 – were distributed among the Democratic and Republican candidates in 1928. We do know that men accounted for as many new voters as women in these Midwestern states (perhaps more in Illinois). And we know that the relatively larger Progressive support among men in these states in these states, documented in Chapter 6, had consequences beyond the 1924 election: It created a pool of men available for conversion in 1928 that far exceeded the number of women so available. Thus women’s greater loyalty to the locally dominant party in 1924 – evidenced by larger Republican vote share and (surprisingly) lesser enthusiasm for the Progressive Party ticket – translated into greater relative loyalty to the traditionally dominant party in the election of 1928, in which women swing to the Democrats to a considerably lesser extent than men in these long-standing Republican strongholds.

In Missouri and Kansas, states with smaller immigrant populations and smaller Progressive vote share in 1924, the patterns are different, and the relative contributions of men and women are generally similar. Consistent with claims that rural, native-born women in the Midwest supported Hoover in opposition to Smith, we find relatively small (or nonexistent) Democratic gains, and moderate (but larger) Republican gains (see Figures 7.11 and 7.12), among women in 1928. However, women are not distinctive; the same pattern characterizes male voters, with the result that net Republican gains (Figure 7.13) are about the same for men and women in both states.

Summary. There was a great deal of electoral disruption in 1928 with clear regional patterns. In the one-party Democratic South and Border states, there was little change in Democratic votes cast among men and women, but a gain in Republican mobilization, and more so for men than for women (we are confident of this finding in Virginia, moderately confident in Oklahoma). As a result, both men and women contributed to the Republican swing in the South, but men more than women. The previously Republican-dominated Northeast experienced a huge Democratic mobilization. Conversion played a larger role among men than among women, but new male mobilization was also sizeable. Democrats netted considerably more votes from men than from women in the Northeast. Parts of the one-party Republican Midwest (Minnesota, Iowa, and Illinois) swung to the Democrats in 1928 (although not enough to change election outcomes) and Democrats netted significantly more new votes from men than from women. This gender difference was apparently a result of the greater numbers of available Progressive voters among men compared to women, as the numbers of new male and female voters (also considerable in most states) were quite similar. In other parts of the Midwest (Republican-leaning Missouri and one-party Republican Kansas) the Republicans netted more votes than Democrats, and to a fairly similar extent from men and women.

Thus, in general, despite the enthusiasm of new women voters for Republicans in states that moved Republican (Oklahoma and Virginia) and for Democrats in states that moved Democratic (Massachusetts and Connecticut, Minnesota and Illinois), men accounted for more of the surge in support for the advantaged party. Clearly, our estimates provide no support for popular scholarly narratives that give women substantial credit for electoral shifts in 1928. Given the degree to which women continued to lag men in turnout, we should not be entirely surprised to find men accounting for a greater portion of electoral volatility in 1928.

Conclusion

Both accounts outlined at the beginning of the chapter are in some ways vindicated by our estimates – the Smith candidacy seems to have mobilized women as Democrats in the Northeast and mobilized women as Republicans in the South. Patterns in the Midwest are less straightforward, with Republicans netting female votes in Missouri and Kansas, but Democrats netting more female votes than did Republicans in Minnesota, Iowa, and Illinois. While we cannot definitively attribute these different patterns to variation in nativity, we do note that the two groups of Midwestern states are distinguished on that dimension, with relatively large first- and second-generation immigrant communities in the Democratic-swinging Midwestern states, and smaller immigrant communities in Republican-swinging Missouri and Kansas.

The starkly different political behavior of women across regions and states – and the general similarity of the behavior of women and men within those same regions and states – reinforces the conclusion that women were subjected to and responded to political cues and messages not as women per se but as localized political actors. The results suggest it is folly to make general claims about women as predisposed in general to support one major party or the other – as is the case with men, in some areas women are overwhelmingly Republican, in other places overwhelmingly Democrats.

Our results for 1928 also challenge attempts to generalize about the party loyalty of women as a group. Women in the Republican-dominated Midwest were drawn in disproportionate numbers to the dominant major party in the state in the first elections after suffrage, but that pattern begins to breakdown in 1928. More importantly, we find evidence of a greater shift away from the locally dominant Democrats among women in Southern and Border states in reaction against Al Smith. Where women in the Midwest maintained their support for the dominant Republican Party, even when offered a Progressive option in 1924, women in the Democratically dominated South and Border states may have been more likely to come out for Republican Hoover in opposition to Al Smith in 1928. Women’s loyalty to the major parties – like that of men – appears to be a function of the conditions and contexts of specific election contexts and not an inherent quality of their gender.

Many contemporary observers and later scholars expected that the greater number of inactive women combined with the heightened passions of the 1928 campaign would provide a large pool of new female voters in 1928 (e.g., Burnham 1974; Sundquist 1983). There is good evidence in our sample that women were stimulated by the election of 1928: In our ten sample states, more than twice as many women were added to the polls in 1928 (more than 924,000) than in 1924 (almost 430,000). But women were not alone in being mobilized by the election of 1928. There were only about 290,000 more male voters in our sample states in 1924, but more than 1 million in 1928! Indeed, of the five presidential elections we examine, 1928 is the only one in which there are not substantially more new female voters than new male voters. The drama of the 1928 election was not distinctly mobilizing for women, but brought both sexes to the polls in record numbers.

Many also expected that large numbers of new female voters were a major contributor to electoral change during this period. Women had had fewer opportunities than men to reinforce their partisanship through repeated voting, resulting in what Converse (1972, 277) has called “the temporary but massive dilution of the overall ‘political socialization’ of the electorate represented by the opening of the system to participation by women.” Converse and others expected that, among other things, the impact of women’s enfranchisement was considerable partisan volatility. Yet, we find that the main impact of women on the dramatic political shifts in 1928 appears to be the dampening of those partisan swings. In states that swung toward the Democratic Party, Democrats netted fewer votes from women than they did from men. In states that experienced a Republican surge, more of that surge was attributable to men’s ballots than to women’s. Much of that change appears attributable to the more numerous male Progressive voters from 1924. Most states witnessed an approximately equal number of new male and female ballots (remarkable, given how many more women were available for mobilization) but greater net Republican gains or losses (Figure 7.13) attributable to men. More conversion among men, particularly from those men who cast Progressive ballots in 1924, appears to explain the difference. Thus, women did not contribute to electoral change in 1928 to a greater degree than men did; if anything, women’s presence stabilized the partisan electorate. In this sense, we agree (thus far) with Burnham (1974, 1015), who argued in direct opposition to Converse and others that the impact of this particular statutory change – female enfranchisement – appeared to be “declines, not increases” (italics original) in electoral volatility.

As many scholars have noted, the partisan shifts witnessed in 1928 – the breakdown of a highly resilient electoral pattern – might not have been maintained beyond the unique Hoover–Smith contest had the Great Depression and related economic and political upheaval not ushered in a permanent political transformation. In Chapter 8 we turn to the momentous elections of 1932 and 1936 in which Republican ascendancy gave way to the dominant Democratic New Deal coalition. What role did female voters – now with more than a decade of experience as enfranchised citizens – play in this most consequential of electoral transformations?