Congressional Parties and Civil Rights Politics from 1933-1972

Eric Schickler UC Berkeley Kathryn Pearson University of Minnesota Brian Feinstein Harvard University

Prepared for the History of Congress Conference, George Washington University, May 29-June 1, 2008.

When a Republican Congress pushed for enactment of legislation to protect African Americans’ voting rights in 1890, the opposition from both northern and southern Democrats was virtually universal, unmatched in its intensity, and ultimately killed the bill after a prolonged Senate filibuster. Thirty years later, when an anti-lynching bill passed the House – only to be scuttled by another Senate filibuster – the coalitional alignment was similar. Just eight Democrats supported the bill when it came to the House floor (as compared to 103 Democratic nay votes), while 221 of 238 House Republicans voted in favor.1 Four decades later, when the landmark civil rights legislation of the 1960s came to the House and Senate floor, the parties’ roles had been partly reversed. Now, northern Democrats were the strongest supporters of civil rights, while Republicans were split and southern Democrats continued to be opposed. Over the next decade, Democrats became identified as the “party of civil rights” while Republicans increasingly became identified with racial conservatism. The reversal in the parties’ positions on civil rights is widely viewed as one of the most important political transformations in recent decades. Yet the timing and origins of this change remain in dispute. Carmines and Stimson (1989) argue that the critical shift occurred in 1963-64, with rapid changes set in motion by Lyndon Johnson, Barry Goldwater and other elite actors. The Democratic majority in Congress that emerged with the New Deal is often depicted as an alliance of conservative southern Democrats and a more liberal northern contingent that was nonetheless largely content to ignore civil rights issues that threatened to split the party. While a handful of northern Democrats – particularly high-profile 1948 Senate victors Hubert Humphrey (D-MN) and Paul Douglas (D-IL) – were clearly committed civil rights activists, most northern Democrats are generally portrayed as indifferent to civil rights. If one takes second dimension 1

The handful of northern Democrats in the House split evenly on the bill: seven voted in favor, six voted nay, and nine abstained. Thus, even northern Democrats were much less supportive of the bill than were northern Republicans.

1

NOMINATE scores as a measure of civil rights liberalism during this period (see Poole and Rosenthal 1997, 109-11, 230), northern Democrats consistently vote more conservatively than do Republicans in the 1930s-50s (see below). By contrast, we argue that the partisan shift on civil rights both began far earlier and proceeded far more gradually than Carmines and Stimson’s (1989) account claims, with northern Democrats’ lead on civil rights making its first appearance during the early-to-mid-1940s, and growing gradually during subsequent decades. We further argue that congressional Democrats’ early support for civil rights measures can be traced to prodding from the party’s core coalitional allies. During the New Deal period, the Democratic Party outside of the South was transformed into the party of CIO unionists, African Americans, Jews, and liberal intellectuals, initially for reasons having little to do with race. As these groups’ civil rights advocacy increased, congressional Democrats were spurred to support civil rights measures. Rather than unconstrained partisan elites shaping their parties’ civil rights positions during the mid-1960s, New Deal coalitional partners encouraged congressional Democrats to support civil rights decades earlier, while Republican members, without these groups under their party’s umbrella, did not face similar pressures. Coalition group pressure did not cause this partisan shift overnight. Rather, the process unfolded incrementally, as groups that became affiliated with the Democratic Party during the New Deal increased their civil rights advocacy, and these groups, in turn, gradually reoriented the Democratic Party to their position.2 Drawing upon new measures of members’ civil rights preferences, our analysis suggests that the break-up of the Democratic north-south coalition was rooted in deep structural forces in the party that had taken root during the New Deal and that were evident in Congress by the end

2

See Karol (forthcoming) for a broader account of the relationship between the parties’ choice of issue stances and coalition group alignments.

2

of World War II. Within Congress, northern Democrats – who generally constituted a majority of their party – had demonstrated a stronger commitment to civil rights than their Republican counterparts long before the 1960s. While Democratic committee and party leaders – many of whom hailed from the South – used their prerogatives to keep civil rights off of the agenda in order to prevent a complete party break, we find that northern Democrats took the lead role in attempting to overcome this gatekeeping. Indeed, southern Democrats found their main allies in resisting efforts to force action on civil rights among Republicans, who showed much less inclination to sign civil rights discharge petitions than did northern Democrats. Our analysis is structured as follows. We first summarize Carmines and Stimson’s issue evolution model, identifying the limits of their reliance on the roll call record as a gauge of congressional preferences. We then describe three alternative preference measures – support for discharging civil rights bills from committee, pro-civil rights rhetoric on the House floor, and sponsoring civil rights legislation – and present our findings. We contrast our results with those based on roll call analysis, and argue that our measures have important advantages in understanding the coalitional alignment on civil rights issues. We then expand on our coalitional partners theory, linking congressional Democrats’ behavior on civil rights issues to pressure from key Democratic constituencies. The conclusion notes several possible extensions of the analysis.

THE ISSUE EVOLUTION MODEL AND CIVIL RIGHTS Among political scientists, Carmines and Stimson’s (1989) issue evolution model is arguably the most widely accepted explanation for the parties’ reversal of civil rights positions during the mid-twentieth century. Two theoretical claims underlie their argument. First, they

3

state that a dynamic growth model best explains the partisan shift on civil rights issues. In this model, most of the change occurs during a relatively brief “critical moment,” which is followed by a longer period of slower, more gradual dynamic growth, during which the political system fully transitions to the new steady state. Applying this model to the partisan shift on civil rights issues, Carmines and Stimson argue that the critical moment occurred in 1963-64. According to their account, both parties took moderate stances on civil rights during the 1940s, 1950s, and early 1960s, with the GOP typically being somewhat more supportive of civil rights measures. Carmines and Stimson begin their study in 1945, when “the politics of race differed … from that at the end of Reconstruction, but not by much” (Carmines and Stimson 1989, 62). For Republicans at that time, there were few negative consequences associated with position-taking on civil rights issues, because civil rights was seen as a “southern issue,” and Republicans’ electoral prospects could not sink much lower in the South. By contrast, the Democratic Party, which was comprised of southern segregationists and go-along-to-get-along northerners, appeared to have a tacit understanding that racial issues not be broached. According to Carmines and Stimson (1989, 62), the party’s otherwise disparate southern and nonsouthern wings had a common interest in advancing the New Deal state, which neither side wanted to jeopardize with an internecine feud over race. They note that at the end of World War II, congressional “urban Democrats managed to avoid conflict between social welfare and race by ignoring race … Notably absent in this picture are the northern Democratic liberals who later became so important to racial politics.” Therefore, when Harry Truman courted Congress to advance his civil rights proposals, he was forced “to turn mainly to the opposition party for support” (62). More generally, they claim that “both

4

parties took relatively moderate stands on racial issues” during the 1950s, although the Republicans tended to be “more supportive of civil rights” (35). Key to Carmines and Stimson’s argument is that the decisive partisan shift on civil rights issues occurred during a brief “critical moment” in 1963-64. Civil rights issues became more salient than in years past, and presidential candidates Lyndon Johnson (D) and Barry Goldwater (R) articulated distinct positions, placing their respective parties on opposite sides of civil rights issues. The 1958 and 1964 congressional elections, which resulted in the replacement of racially liberal Republicans with racially liberal Democrats, facilitated this partisan shift on civil rights (Carmines and Stimson 1989, 63). According to Carmines and Stimson’s account, the partisan shift on civil rights was instigated by national elites. This argument comports with their broader theory of issue evolution, which rests in part on a specific sequence (Carmines and Stimson 1989; Carmines and Wagner 2006). First, the two parties’ elites take clear and distinct positions on an issue. Activists then receive these new positions from elites and relay them to the mass public.3 Finally, voters perceive the two parties’ new positions on the issue and alter their voting behavior accordingly. In this elites-to-activists-to-masses sequence, elites are not merely the first mover – they also have broad discretion in deciding which policy positions they introduce (Carmines and Stimson 1989, 179). Barry Goldwater, for instance, is seen as particularly important in the Republican Party’s evolution on race (1989, 57). Carmines and Stimson emphasize that national elites enjoy broad latitude in crafting their parties’ issue positions.4 3

The term “party elites” includes the president, members of Congress, and candidates for high office; the “activists” label applies to party convention delegates, minor officeholders and party officials (Carmines and Stimson 1989, 162; Carmines and Wagner 2006; Carmines and Woods 2002). 4 This is not to say that elites “control” the process entirely: since they have incomplete information, they cannot predict which policy positions will resonate with voters. Therefore, even though they have great discretion over which positions they adopt, there is a degree of randomness in their early-stage decisions (Carmines and Stimson 1989, 18, 179).

5

An analysis of congressional behavior is a key component of Carmines and Stimson’s (1989) test for their account of the civil rights issue evolution. Their analysis of congressional leadership on civil rights, however, relies exclusively on roll call floor votes. They find that, in the aggregate, House Republicans were consistently more supportive of civil rights legislation than their Democratic colleagues were between 1945 and the early 1960s. Although the differences between the parties’ roll call records begin to narrow in approximately 1962, the Democrats overtake the Republicans as the more pro-civil rights party in 1964. Much the same dynamic occurs in the Senate. Carmines and Stimson conclude that Republicans were greater advocates of civil rights through the late 1950s, when “Republican senators were more liberal overall … and they were also more liberal than the Democrats within each region” (Carmines and Stimson 1989, 70).5 They stress that northern Republicans were at least as supportive of civil rights as were northern Democrats, and the two groups’ similar positioning is what makes it plausible that either party (or both parties) would embrace racial liberalism when the civil rights struggle culminated in the mid-1960s. In recent years, cracks have emerged in the academic consensus on the partisan shift on civil rights, as a group of scholars has cast doubt on some of the issue evolution model’s key claims. For instance, Anthony Chen and David Karol show that, in statehouses outside of the South and in Congress, respectively, the Democrats’ emergence as the more receptive party on civil rights issue is evident far earlier than Carmines and Stimson’s 1963-64 critical moment. Chen (2006, 2007) argues that Republican opposition to state-level civil rights measures took root in the mid-1940s. Karol (1999; forthcoming), examining congressional roll calls, finds that nonsouthern Democrats emerged as slightly more liberal than Republicans on floor voting during

5

The authors acknowledge that outside of the South, the difference of means between the two parties’ Senate delegations is not statistically significant, although its direction is supportive of their claim.

6

the 1940s (although these differences do not become sizeable until the 1960s). Lee (2002) questions the notion that party elites drove the civil rights shift. Rather than following an elitesto-activists-to-masses channel, Lee argues that a broadly-based social movement mobilized mass opinion on civil rights, challenging party elites. Feinstein and Schickler (2008) challenge both the timing and sequence components of the issue evolution model, using a large database of state political party platforms to demonstrate that state-level Democrats took the lead on civil rights during the mid-1940s across a wide range of nonsouthern states. Feinstein and Schickler suggest that Democrats’ key New Deal coalitional partners – i.e., unions, ethnic and religious organizations, and progressive policy groups – pushed the party towards a pro-civil rights orientation far earlier than Carmines and Stimson claim. In this paper, we turn our attention to the parties in Congress. Whereas past scholarship has relied on the roll call record as a measure of members’ preferences, we examine members’ signatures on discharge petitions, speeches on the House floor, and bill sponsorship, which we argue provide a more complete picture. We show that northern Democrats’ lead on civil rights began much earlier and grew more gradually than the issue evolution model suggests. TOWARDS A BETTER UNDERSTANDING OF SUPPORT FOR CIVIL RIGHTS Roll Call Analysis of Civil Rights Before presenting our alternative measures of coalitional alignments on civil rights, we briefly turn to the roll call record to summarize what it reveals and highlight its limitations. Perhaps the most commonly used measures of party positioning are NOMINATE scores. Figure 1, which is reproduced from McCarty, Poole, and Rosenthal (2007), shows that on what McCarty, Poole, and Rosenthal label the “civil rights dimension,” Republicans are to the left of northern Democrats throughout the period we examine. While the second dimension is arguably no

7

longer significant – or focused on civil rights – by the 1970s-80s, the pattern is consistent with the conventional wisdom that Republicans were more inclined to support civil rights initiatives than northern Democrats in the 1940s-50s. Figure 1

This figure comes from: http://www.voteview.com/polarized_america.htm.

Rather than relying on NOMINATE scores – which are based on all roll calls cast in a given Congress -- Carmines and Stimson construct civil rights support scales based on selected roll calls in each Congress. As noted above, they find that Republicans are to the left of Democrats on civil rights throughout the 1930s-early 1960s, and that this basic pattern holds

8

within each region (though the differences are much smaller and not necessarily statistically significant). We have expanded upon Carmines and Stimson’s analysis by modeling each civil rights roll call vote separately to determine whether nonsouthern Democrats or nonsouthern Republicans were more likely to support civil rights.6 Figure 2 presents simulated first differences for a model that estimates the likelihood of a pro-civil rights vote in a simple logit model, controlling for seniority.7 The results differ only marginally from Carmines and Stimson’s, and tell a similar story to the support scales constructed by Karol (forthcoming). First, nonsouthern Democrats and Republicans vote similarly on nearly all of the roll calls. While the first differences tend to suggest that starting in the mid-1940s, northern Democrats are a bit more likely to cast yea votes than are northern Republicans, the point estimates are quite small, indicating that the gap in the probability of a pro-civil rights vote between the two groups is typically in the .05 to .10 range.8 Thus, it is easy to see why one could conclude that northern Democrats were not stronger supporters of civil rights than were northern Republicans if one focused solely upon roll call votes.

6

We thank John Lapinski for providing the roll call codes. The results thus far are preliminary. There are a few additional roll calls from the Lapinski dataset that we plan to analyze in the future. 7 Seniority is held at its mean in computing the first differences. All figures in this paper involving differences in expected values were obtained using a logit model. We run 1000 simulations in all analyses, using the CLARIFY software package (Tomz, Wittenberg and King, 2003). 8 Similarly, in the 1930s, when Republicans were more pro-civil rights than Democrats, the gap evident in the discharge petition analysis is much more substantial than in the roll call votes.

9

Figure 2

Two apparent exceptions are worth noting. A 1937 anti-lynching roll call seems to show northern Democrats as much stronger supporters than were Republicans. However, this was because the bill under consideration was widely viewed as weak; Republicans thus voted against the bill in order to pave the way for consideration – a few days later – of a stronger bill. Southern Democrats joined the Republicans in voting nay, but this was due to their opposition to

10

any civil rights legislation. Northern Democrats, however, were content to support the weaker bill.9 Thus, this example reinforces the conclusion that Republicans were more supportive of civil rights than were Democrats in 1937. Indeed, when the stronger bill came up later in that Congress, the model indicates Republicans were more supportive than northern Democrats (though the first difference estimate is in the .10 range). The second exception is the Powell amendment of 1956. In that case, Republicans took the pro-civil rights position at a much higher rate than northern Democrats. As the voluminous literature on the Powell amendment makes clear, strategic calculations were at work: northern Democrats were reluctant to vote for the amendment because it threatened to sink aid to education legislation.

The Limits of Roll Call Votes as an Indicator of Civil Rights Preferences If one only examines roll call votes, it seems reasonable to argue that either party could have embraced civil rights liberalism in the mid-1960s. After all, the gap separating nonsouthern Democrats from nonsouthern Republicans was quite small, and the Democrats had to contend with their anti-civil rights southern wing. But the roll call evidence belies a much more substantial gap between nonsouthern Democrats and Republicans. Roll call votes are a problematic measure of member preferences with regard to civil rights issues.10 Only a small fraction of items on the congressional agenda reach the floor for a roll call vote. Civil rights measures were particularly prone to gatekeeping; indeed, in several Congresses, there were few roll call votes on civil rights despite its prominence on the committee

9

See “House Prepares for Passage of Anti-Lynching Bill,” Washington Post, April 8, 1937; and “House Sidetracks Anti-Lynching Bill,” New York Times, April 8, 1937. Any doubt about the strategic situation is resolved by noting that the bill (which northern Democrats voted to consider while Republicans opposed consideration) was called up by Judiciary Committee Chairman Hatton Sumners of Texas, who was a foe of civil rights. 10 See Clinton and Lapinski (2007) for a more general assessment of the limits of roll calls as a lens to understand lawmaking activity.

11

agenda. For example, there were just 19 civil rights roll call votes on the House floor from 193348, and a handful of those votes incorporated other issues and thus are not particularly good indicators of civil rights support.11 Democrats’ reliance on the seniority system to select committee chairs, coupled with southern Democrats’ disproportionate levels of seniority in the chamber (due to their election from safe districts), meant that staunch civil rights opponents often chaired the House committees with jurisdiction over civil rights (Nicol 1994). The Judiciary Committee, chaired by Hatton Sumners (D-TX), for example, was a “graveyard” for civil rights bills from 1932 to 1947.12 According to Bensel (1984), no major civil rights bill was defeated in a straight up-ordown vote on the House floor in the 1920s-60s. Instead, legislative failures have “come at the hands of the committee system” and Senate filibusters (1984, 234). Southern Democrats hardly acted alone to prevent civil rights legislation from reaching the House floor. An alliance of Republicans and southern Democrats on the Rules Committee forged an additional roadblock to the consideration of civil rights legislation. Inaction by the Rules Committee killed many bills during this era. The Rules Committee also prevented the consideration of liberal amendments; ensured that conservative-backed amendments received favorable terms; and required significant changes in bills as a precondition for reporting them to the floor. Unsurprisingly, this cross-party conservative alliance on the Rules Committee was particularly hostile to civil rights measures. Rules Committee members consistently lagged behind their House colleagues in their support for civil rights bills on the floor in all but one year between 1922 and 1965. The size of this gap between committee Democrats’ support for civil

11

We thank John Lapinski for providing a comprehensive list of civil rights roll calls from 1930-70. See Katznelson and Lapinski (2006) on the coding scheme. 12 In fact, according to Bensel (1984, 238), during the 74th Congress alone, “the Judiciary Committee brooded silently over 32 anti-lynching bills.”

12

rights roll calls and floor Democrats’ support peaked in 1937, when Bensel (1984, 239) estimates that approximately 60 percent of House Democrats favored enactment of an anti-lynching bill, while only one-third of the Democratic caucus on the Rules Committee favored this measure. Southern Democrats and Republicans on the Rules Committee frequently used their levers of influence to block civil rights measures from receiving a vote, while facilitating floor consideration of conservative bills (Lapham 1954; Schickler 2001; Schickler and Pearson 2007). Only by circumventing the Rules Committee did civil rights legislation typically reach the floor for a vote between 1933 and 1965. Four major civil rights bills were discharged from Rules during this period, the committee was bypassed via suspension procedures on three occasions, and in two other instances the committee reported a major civil rights bill to the floor only under a discharge threat (Bensel 1984, 239).13 Critics at the time speculated that Republicans were cooperating with the southern Democrats to keep civil rights off the floor in exchange for the southerners help on economic legislation.14

Three New Measures of Civil Rights Preferences The importance of gatekeeping suggests not only that the roll call record will be highly censored, but that the key stage in pushing civil rights bills to House passage involves circulating and signing discharge petitions to overcome negative agenda control. While roll call votes were a blunt instrument for distinguishing among nonsouthern members’ commitment to civil rights liberalism, we believe that signatures on discharge petitions offer a more refined measure. We also consider two additional measures: members’ speeches on the House floor and their

13

By “discharge threat” we mean that the number of signatures on a discharge petition is nearing 218. See MacNeil 1963; Kesselman 1948; Berman 1962, 73; and Evans and Novak 1966, 30. Other authors argue that the GOP-southern Democratic coalition was rooted in shared policy goals, rather than a logroll (see, e.g., Brady and Bullock 1980).

14

13

sponsorship of civil rights legislation. Unlike roll call votes, the indicators derived from discharge petitions, floor speeches, and bill sponsorships are not censored by negative agenda control. These new measures allow us to begin to test our hypothesis that nonsouthern Democrats’ embrace of civil rights liberalism was rooted in deep currents within the party that emerged in the 1940s. It is worth noting that these measures tap two elements of member preferences with respect to civil rights policy: a member’s ideal point (i.e., the most preferred policy) and the intensity of a member’s preference (i.e., how much the member cares about implementing that ideal point relative to other priorities and issues). Roll call-based measures are generally good at capturing ideal points – so long as there are a sufficient number of roll calls with a variety of cutting lines.15 But for understanding changes in coalitional alignments over time, intensity is also important. Even if Republicans and northern Democrats have similar ideal points on civil rights policy in the 1940s-50s (at least as reflected in their votes on the limited number of roll calls that reach the floor), it matters whether the groups differ in how much they care about the issue. Thus, if northern Democrats are intense supporters while most Republicans care little about the issue, it should be no surprise that Republicans later proved willing to make common cause with southern civil rights opponents in the 1960s when the increased salience of the issue paved the way for GOP gains in the region. The measures we employ capture both the ideal point and intensity facets of preferences, and thus are a useful window for understanding the changing coalitional alignment on civil rights from the 1930s-60s.16 15

As noted above, the small number of civil rights roll calls in many Congresses creates difficulties for estimating ideal points on this issue. 16 Many observers have commented on differences among members in preference intensity for civil rights and how these differential commitment levels can affect whether legislation is enacted. For instance, Rep. Richard Bolling (D-MO) believed that differences in preference intensity on civil rights issues were large enough to allow for categorization, estimating that approximately 50 Republican members and twice as many Democrats were truly concerned with civil rights in 1956 (Bolling 1965, 175). Bensel (1984, 233) also notes the importance of differences

14

Discharge Petitions Discharge petitions provide members the opportunity to bring legislation with simple majority support to the House floor that has been bottled up by committees or party leaders. Since 1931, members have had the ability to discharge substantive legislation and special rules from committees. A member of Congress may file a discharge petition once a bill or resolution has been stuck in committee for 20 days, or 7 days in the Rules Committee (Beth 2003). Once the petition has 218 signatures, the signatories’ names are entered into the Congressional Record, and after a motion to discharge passes on the floor, the committee under question is discharged.17 Determining which members use the discharge process to support civil rights legislation provides new insight into the partisan dynamics of civil rights legislation. Discharge petitions reveal that a group of members cared enough about an issue to mount a drive to bypass the usual legislative process. As a preference measure, discharge petitions differ from roll call votes in two important ways. First, a member’s ability to sign a petition is not constrained by negative agenda control. Second, discharge petitions reveal the bills that a subset of members cared most about that were being blocked from consideration on the House floor, and by extension, which members were most dissatisfied that civil rights bills were being blocked from consideration. In addition, discharge petitions could be consequential. They brought at least four civil rights measures directly to the floor and likely influenced the consideration of several others. A 1937 anti-lynching bill and anti-poll tax measures introduced in 1942, 1943, and 1945 came to the House floor for roll call votes only after successful discharge petitions dislodged them from in preference intensity – and points to discharge petitions, floor speeches, and bill sponsorship as a measure of priorities. 17 The rules governing discharge petitions have remained relatively constant during the period under study (the 73rd92nd Congresses, 1933-1976), with one exception: during the 73rd Congress, only 145 signatures were necessary to discharge a bill; that threshold has been set at 218 in all subsequent Congresses (Beth 2003).

15

the Judiciary and Rules Committees. In addition, committees may have reported civil rights bills to the floor under threat of a growing list of signatures on a discharge petition. For example, the Rules Committee, mindful of discharge petitions that were approaching the 218 signature threshold, reluctantly reported a special rule on the landmark Civil Rights Acts of 1960 and 1964. Had this discharge threat not existed, it is unlikely that the Rules Committee would have acted (Bensel 1984, 236). This is not to say, however, that signing a discharge petition is a costless action. Most obviously, seeking out the petition to sign and encouraging one’s colleagues to do the same takes time. Some members view signing a discharge petition as violating congressional norms and intruding on committee authority (Oleszek 2004), potentially drawing the ire of chairs and other influential members. Along these lines, it is worth noting that other members could view the list of signatories, even as the names were not made public. These costs offer a potential explanation of why, out of the thousands of bills that die in committee during each Congress, discharge attempts are undertaken for only a small fraction. Given these costs, signatures on these petitions signal preference intensity. If members care enough about a bill to seek out a petition to sign and are willing to absorb the potential negative consequences associated with violating committee norms, that action typically carries more information than casting a roll call vote does. Thus, whereas both roll call votes and discharge petition signatures signal preferences, signing a petition also signals priorities. Despite their promise as an indicator of preferences, research on the discharge procedure has been limited. Until 1993, the names of the members who sign the petitions were only made public in the 48 cases in which the petition reached the threshold number of signatures and was

16

printed in the Congressional Record.18 However, recently discovered heretofore unexamined discharge petitions filed from the 71st through the 94th Congress at the National Archives has made the analysis of discharge petitions possible (Pearson and Schickler 2007). There were 34 discharge petitions filed from 1933-76 on pro-civil rights bills; there also were a handful of additional petitions for bills to roll back civil rights initiatives – particularly busing – starting in the early 1970s. We have constructed a dataset that includes the signatories on each discharge petition, along with the order in which the members signed and information about each member’s party, state and district, ideology, committee assignments, and seniority. We restrict the analysis to the 29 pro-civil rights petitions that received twenty or more signatures. The procivil rights petitions involve anti-lynching, fair employment practices, anti-poll tax, and other civil rights legislation. For comparison, we also include the anti-busing discharge petitions that were circulated in 1971-76. Of the 29 pro-civil rights discharge petitions, seven reached the signature threshold needed to discharge the bill from committee. Since members could generally expect that discharge petition signatures would only be made public if the petition reaches 218 signatures, we believe that the decision to sign – particularly for members who sign well before the petition nears the threshold – is an indicator of the signer’s sincere preferences. We present some initial evidence on the sequencing of signatures below, but plan a more complete analysis in future drafts. The only systematic analysis of civil rights discharge petitions that we have found is presented in Bensel’s (1984) classic Sectionalism and American Political Development: 18801980, which argues that there was an “ideologically committed” element in the northern wing of the Democratic Party that hoped to force civil rights onto the agenda, but that many other

18

During the time period we investigate, the threshold has been 218 signatures in all congresses except for the 72nd and 73rd, when discharge petitions only needed 145 signatures.

17

northern Democrats were moderates who preferred that the divisive issue be shelved in order to preserve the coalition with the South. Bensel argues that party leaders and the northern moderates used the committee system to protect the party by preventing civil rights from coming to the floor, and that only Democrats representing “industrial labor constituencies seemed sincerely enthusiastic” about enacting civil rights legislation, while other northern Democrats were ambivalent (Bensel 1984, 232). He points specifically to machine Democrats’ disinclination to sign discharge petitions as evidence that their public support for civil rights was not entirely sincere (233). Bensel analyzes two specific discharge petitions – the successful 1937 petition to circumvent obstruction of an anti-lynching bill and the petition for the Civil Rights Act of 196019 – and shows that on both occasions, committee chairmen and members of the discharged committee were less likely to sign than were their fellow partisans. However, in 1960, Republicans as a whole were less willing to undermine committee gatekeeping and sign the discharge petition than in 1937, while Democrats became slightly more likely to sign in the later case. Bensel’s account suggests that the change between 1937 and 1960 occurred late in the period and was relatively tentative. From these two petitions, Bensel concludes that northern core Democrats exhibited a greater willingness to take on the southerners by 1960, but that it was not until 1965 that a “decisive and persistent” switch in party alignments took place (p. 242). Our complete petitions dataset allows for a more fine-grained analysis.20

19

In the latter case, the petition did not reach 218 signatures but the names of 209 of the 211 signatories were leaked to the press. It was extremely unusual for the names of signatories to be leaked to the press since this violated House precedent (see Pearson and Schickler 2007). However, this tactic was used by civil rights advocates on at least two occasions. In both cases, civil rights supporters evidently leaked the petition in order to force the hand of Republicans who claimed to support civil rights but had failed to sign. 20 Bensel interprets the reluctance of members of the targeted committee and of committee chairmen to be a sign that such members were defending the Democratic Party’s institutional interest in keeping civil rights off the agenda. However, Pearson and Schickler (2007) show that target committee members and committee leaders were more likely to oppose discharge efforts regardless of the issue during the 1930s-70s. An extension of the present analysis will be to examine whether the impact of these variables on civil rights petitions declines as the north-south coalition breaks down.

18

We use these new discharge petitions signatures data to determine the relative likelihood of nonsouthern Democrats and nonsouthern Republicans signing civil rights-related discharge petitions. We analyze each petition separately. Figure 3 presents first differences in the expected probability of Democrats and Republicans signing each civil rights-related discharge petitions, setting other potentially relevant covariates at their average values and excluding all southern members. Positive values indicate that nonsouthern Democrats were more likely to sign than nonsouthern Republicans. As Figure 3 illustrates, while Republicans were more willing to sign discharge petitions on civil rights issues than were nonsouthern Democrats during the early to mid-1930s, the two groups seemed indistinguishable during the late1930s and early 1940s. More importantly, the nonsouthern Democrats had taken the lead on efforts to move civil rights legislation to the floor by the mid-1940s. While we acknowledge that pinning down exact dates for these three periods is challenging, we argue that the GOP was more supportive during the 73rd-75th Congresses (1933-1938); we observe mixed or null findings regarding which group was more supportive during the 76h-77th Congresses (1939-1942); and the nonsouthern Democrats became the more supportive group starting in the 78thCongress (1943-1944). By the late 1940s, the gap in predicted probabilities was typically in the .40 to .60 range – indicating a vast difference in the propensity of nonsouthern Democrats and Republicans to support efforts to force action on civil rights.21 Unlike the roll call voting analysis, the difference between nonsouthern Democrats and Republicans is stark from the late 1940s onwards.

21

The sole small point estimate after 1945 occurred in the 81st Congress on a bill that garnered just 24 signatures (even in that case, 19 of the signatures were from northern Democrats).

19

Figure 3

20

It is worth emphasizing that the pattern is not attributable to changes in the substance of the issues before Congress. For example, Republicans were more likely than nonsouthern Democrats to sign discharge petitions for the anti-lynching bills in the 74th-75th Congresses.22 But in the 78th Congress (1943-44), nonsouthern Democrats were significantly more likely to sign the petition for the anti-lynching bill – and this difference persisted on the anti-lynching petitions in the 79th and 80th Congresses. One sees a similar pattern on poll tax bills: Republicans were only slightly less likely to sign petitions to bring a poll tax bill to the floor in the transitional 76th Congress than were nonsouthern Democrats,23 but the gap increased noticeably by the 78th-79th Congresses. This finding presents two challenges to the issue evolution model. First, our results indicate that the partisan shift on civil rights occurred far earlier the 1963-64 point at which Carmines and Stimson (1989) claim that “first mover” partisan elites (members of Congress, along with the president and candidates for high office) shifted positions on civil rights. Instead, the first period for which we observe a clear northern Democratic lean in petition signatories is the 78th Congress (1943-1944) – a full two decades earlier than 1963-64. Second, the “critical moment” label does not apply to any year in our analysis. Whereas Carmines and Stimson claim that much of the partisan shift on civil rights issues occurred during a brief “burst of rapid change,” followed by much lengthier period of secular realignment, we cannot identify any such critical moment – not at 1963-64 or any other years. Rather, we observe a long, slow process, during which the Democrats gradually replace the Republicans as the pro-civil rights party. Nonetheless, it is striking that the magnitude of the gap separating nonsouthern Democrats and Republicans had become substantial by the 81st Congress (1949-50) and remained stable 22

The difference was statistically significant for two of the three relevant petitions in the 74th-75th Congresses. The first difference between nonsouthern Democrats’ and Republicans likelihoods to sign is 0.103, with a 95 percent confidence interval of 0.007 to 0.204.

23

21

thereafter. That is, in terms of discharge petition activity, the realignment of nonsouthern Democrats and Republicans had largely been completed by the early 1950s. We have started to examine the sequence of member signatures on the petitions as well. Thus far, the results tell a similar story to the simpler binary model of who signs each petition. At the start of the period, when Republicans were more pro-civil rights than nonsouthern Democrats, they not only were more likely to sign the petitions, but among the signatories, they were more likely to sign early. By contrast, those Democrats who did sign the petitions tended to do so later in the process. This suggests that the most intense support was concentrated among Republicans. For example, Figure 4 below presents a kernel density plot of the signature sequence for the first petition in our dataset, a resolution to investigate segregation in the House restaurant. Notice that Republicans were more likely to be included among the early signatories. Indeed, 33 of the first 50 signatures were from Republicans as compared to twelve (24%) from nonsouthern Democrats (the remainder were from minor parties). By contrast, nonsouthern Democrats accounted for 44% of the last 50 signatures on the petition.

22

.002

.004

Density .006

.008

.01

Figure 4: Kernel Density Plot of Signature Sequence for 1934 Petition on House Restaurant Segregation

0

50

dp_14

100

150

(Blue: Northern Democrats; Red: Republicans)

For the petitions in the late 1930s-early 1940s, there is no clear pattern as to which group signs earlier. However, by the mid-1940s, those nonsouthern Democrats who sign petitions tend to do so earlier in the process than do the Republican signatories. The first discharge petition in the 79th Congress (1945-46), which concerned a bill to ban the poll tax, underscores the change. As the density plot shows (see Figure 5), nonsouthern Democratic signatures were concentrated early in the process, while Republicans largely waited to sign.24 Northern Democrats accounted for 42 of the first 50 signatures, while Republicans accounted for just six (with one minor party

24

The general pattern holds across the vast majority of civil rights petitions; the size of the difference between the parties varies a bit, but the tenor of the results is consistent.

23

member and one southern Democrat also signing). By contrast, Republicans accounted for 39 of the last 50 signatures – which were what allowed the petition to reach the threshold of 218.

0

.002

Density .004

.006

.008

Figure 5: Kernel Density Plot of Signature Sequence for 1945 Petition on Poll Tax

0

50

100

150

200

250

(Blue: Northern Democrats; Red: Republicans)

In a 1959 speech entitled “The Civil Rights Discharge Petition: A Reliable Political Divining Rod,” Rep. William Moorhead (D-PA) claimed that, with both parties offering cheaptalk on civil rights, one ought to examine discharge petitions as an indicator of the parties differing levels of support for civil rights initiatives.25 Our exploration of this question provides an emphatic answer to Moorhead’s challenge. Nonsouthern Democrats had become the most

25

Congressional Record, 86th Congress, 1st Session, Vol. 106, Part 1, p. 1205.

24

vigorous supporters of civil rights legislation by the mid-1940s. Our evidence shows that Republicans were privately less supportive of civil rights bills than their roll call votes suggest, but does not allow us to determine whether this gap was due to their sincere preferences or part of a deal with the southern Democrats.26 The bottom line, however, is that the GOP’s later embrace of racial conservatism is presaged by their willingness to support southern Democrats in their efforts to keep civil rights initiatives off of the floor starting in the 1940s, and offers a sharp contrast both to the Republicans’ more liberal floor voting record and to nonsouthern Democrats’ more vigorous civil rights advocacy.

Floor Speeches To explore whether our discharge petition results reflect a broader pattern of greater nonsouthern Democratic commitment to civil rights legislation, we examine two additional measures: speeches on the floor and bill sponsorships. When members have the floor, they may expound on whatever subjects they choose. In doing so, they balance competing demands on their time. Therefore, those issues that members choose to devote their attention to on the floor provide some insight into which issues they are most interested – or most desire to convey an interest, at least.27 Given that Carmines and Stimson prioritize the signals sent by elites in their account of the shift in support for civil rights, members’ own words should be one of the first places to look for such cues. In an effort to determine the extent to which Democratic and Republican members are vocal supporters of civil rights initiative, we searched the Congressional Record, coding each

26

We are eager to hear suggestions on how to tease this out. Our use of speechmaking as a measure of priorities is consistent with Hall’s (1996, 3) view that differential participation in various committee activities provides an indicator of “revealed intensities.” For a specific application of Hall’s approach to congressional speechmaking, see Cannon (1999, 188).

27

25

member making a floor speech in favor of a civil rights initiative in each Congress.28 During the 73rd-92nd period (1933-1972), the median number of members in each Congress delivering at least one floor speech in favor of civil rights measures was 34.29 Figure 6 shows the increase in the number of members speaking out on civil rights issues during the period under study. The data points and associated lowess regression line show that the number of speakers remains fairly stable until approximately the 81st Congress (1949-1950), when it begins a gradual and steady (in fact, almost monotonic) increase through the end of the period. If floor speeches reflect the salience of an issue among elites, it appears that no one particular Congress stands out as a period of particularly rapid increase in civil rights’ salience.30

28

This procedure creates a simple dichotomous measure indicating whether or not each member spoke out in favor of civil rights measure on the floor in each Congress – it does not record the number of pro-civil rights speeches that the member gave, nor do we attempt a content analysis to rank different members’ civil rights positions. To identify pro-civil rights speeches, we search the Congressional Record in each Congress for the following terms: antilynching, civil rights, colored [citizens, rights, etc.], desegregation, fair employment, fair housing, FEPC, integration, Jim Crow, lynching, Negro [rights, etc.], and segregation. A speech is defined as over four contiguous lines of text delivered by one speaker on any of the above subjects. We exclude all text in the Record that was not delivered orally on the House floor, e.g., supplemental material inserted into the appendix. 29 Each of these speeches was delivered on the House floor – not text inserted in the Congressional Record. 30 Although there is a major increase in speeches between the 87th and 88th Congress (1961-62 and 1963-64) – at Carmines and Stimson’s critical moment – this increase is part of a much longer upward trend.

26

Figure 6

As with our discharge petitions data, we examine the relative likelihood of nonsouthern Democrats and Republicans delivering pro-civil rights speeches on the floor of the House to assess the two groups’ relative support for civil rights measures. The dependent variable is coded as “1” if a member delivered one or more pro-civil rights speeches in a given Congress. We employ statistical simulations to determine the first differences in the expected probability of nonsouthern Democratic and Republican members speaking out on the floor in favor of civil rights in each Congress. Since junior members during this period were less likely to engage in publicity-seeking activity, such as speechmaking on all issues, (Asher 1973), we include a

27

variable indicating seniority, which is held at its mean for this simulation. Figure 7 shows the results of this analysis.

Figure 7

There was no partisan difference in the number of pro-civil rights speeches until approximately the 79th Congress (1945-46), when the nonsouthern Democratic members were slightly more likely to give civil rights speeches than were Republicans.31 In most subsequent Congresses, one observes a gradual increase in the nonsouthern Democrats’ edge over the

31

Although, when one considers how close the lower bounds of the associated 95 percent confidence interval is to zero (i.e., no difference in expected values between the two party delegations), this first difference in the 79th seems even smaller.

28

GOP.32 While there are small dips in some Congresses, the overall trend in Figure 7 is clear: there is a gradually growing gap between Democratic and Republican members’ rhetoric on the House floor regarding civil rights. These results comport with the discharge petitions analysis presented in Figure 3. In both cases, nonsouthern Democrats are more likely to advance a civil rights agenda than nonsouthern Republicans beginning in the mid-1940s, with this gap between the parties gradually increasing in subsequent years.33

Bill Sponsorship Sponsorship of civil rights legislation offers a third test of Democratic and Republican members’ divergent preferences on civil rights. Members may sponsor legislation for a variety of reasons (Schiller 2000), but regardless of their motives, bill sponsorship is a necessary precondition for any legislative action and a public sign of strong support for a particular policy. Examining the changing balance of Democratic-sponsored civil rights bills and Republicansponsored bills allows for greater insight into the timing and sequence of the partisan shift on civil rights issues.

32

The one noticeable “jump” in the Democrats’ relatively greater propensity for civil rights speechmaking occurs between the 89th and 90th Congresses. During the 89th Congress, 68 Democrats and 23 Republicans take the floor; during the 90th Congress, those figures drop to 58 and 10, respectively. In other words, MCs of both parties are less likely to give pro-civil rights speeches in the 90th Congress as compared to the 89th, but this drop is more acute among Republicans. We theorize that two events during the 90th Congress (1967-68) – the riots following the assassination of Martin Luther King, Jr., and House consideration of the Civil Rights Act of 1968 – may explain this large decline in Republicans’ speechmaking in the 90th, relative to the 89th. Many nonsouthern Democrats’ speeches during the 90th called for passing the Act as a tribute to Dr. King, while southern Democrats argues that Congress should not “appease” rioters by passing the bill. These two arguments may have left pro-civil rights Republicans conflicted, given the GOP’s reputation as the “law-and-order” party, and therefore, these crosspressured (pro-civil rights, pro-law enforcement) Republicans may have chosen to remain silent. 33 The main difference between the discharge petitions and floor speeches analyses is that for the discharge petitions, the distance between the two grew until approximately 1950, remaining relatively steady after that year. By contrast, the gap between the two parties continued to steadily increase over time in the floor speeches analysis.

29

Data on bill sponsorship 80th through 92nd Congress (1947-1972) were obtained from Adler and Wilkerson’s Congressional Bills Project website.34 (Regrettably, we were unable to locate data from earlier Congresses, during the 1930s through the mid-1940s.) We coded each bill that Adler and Wilkerson classified as pertaining to “general civil rights,” “ethnic minority and racial group discrimination,” and “voting rights and issues,” to determine whether each bill met our criteria for inclusion.35 We only examine the sponsor of each bill, not its cosponsors. During the 1947-1972 period, 492 distinct civil rights bills were introduced. Figure 8 provides a first look at these data. The figure shows the number of civil rights bills with Democratic sponsors alongside the number of bills with Republican sponsors, along with the associated lowess regression line. As the figure shows, Democrats introduced more civil rights bills than the Republicans in 12 of the 13 Congresses under study; only in the 80th Congress do Republicans introduce more civil rights bills (six versus four bills).

34

Adler and Wilkerson, Congressional Bills Project: 1947-1972. Bills with the expressed purpose of limiting civil rights (e.g., H.R. 383 in the 92nd Congress: a bill “[t]o repeal the Civil Rights Act of 1964”) were excluded, as were bills that focused on expanding civil rights to groups other than racial, ethnic and/or religious minorities (e.g., H.R. 2120 in the 80th Congress, a proposed constitutional amendment lowering the voting age to 18). We focus exclusively on bill sponsorship – and do not consider cosponsorship – because rules governing cosponsorship were in flux during the period under study. The practice of multiple MCs cosponsoring a single bill was prohibited between 1909 and 1967. This ban was relaxed in 1967, when a House resolution permitting up to 25 cosponsors on a bill was passed. This 25 cosponsor limit was lifted in 1978 (Wilson and Young 1997). For “duplicate bills” – i.e., multiple bills with identical descriptions that are introduced in the same Congress – only the first bill introduced is included in these analyses. Including these duplicate bills does not significantly change any of the findings presented in this section, however. 35

30

Figure 8

In addition to the bivariate results presented in Figure 8, we add a control for seniority and conduct a negative binomial regression to analyze bill sponsorship.36 As with the discharge petitions and floor speeches analyses, we employ statistical simulation to determine the first differences in the expected number of civil rights bills that nonsouthern Democratic and nonsouthern Republican members sponsor in each Congress. Figure 9 shows the results of this analysis. 36

We employ a negative binomial model in place of a Poisson model for event count data because the dependent variable (number of civil rights bills sponsored by each MC in each Congress) is overdispersed (mean = 0.085; variance = 0.98).

31

Figure 9

Figure 9 shows that nonsouthern Democrats tended to sponsor more civil rights bills than did nonsouthern Republicans during most Congresses in this period. These results are not as striking as those presented in our discharge petitions and floor speeches analyses. They nonetheless supportour contention that nonsouthern Democrats became the strongest civil rights advocates in the 1940s rather than Carmines and Stimson’s view that northern Republicans were as strong (or stronger) civil rights supporters as northern Democrats prior to the 1960s and that

32

the Democrats became the more pro-civil rights party during a “brief burst of rapid change” centered at 1963-64.37 Reassessing the Issue Evolution Model of Partisan Change The analyses of discharge petitions, floor speeches, and bill sponsorship presented above show that the partisan reorientation on civil rights issues in Congress was an incremental process. The first appearance of a more racially liberal nonsouthern Democratic delegation can be placed at approximately the 78th or 79th Congress (1943-46), and the gap separating nonsouthern Democrats from Republicans became substantial by 1950. This finding is consistent with recent work examining state parties’ civil rights stances. Feinstein and Schickler (2008) examine state party platforms published between 1920 and 1968, finding that state Democratic parties first tend to take more racially liberal stances than their same-state Republican counterparts during the mid-1940s, and the two same-state parties’ positions gradually diverge from each other in subsequent years (Feinstein and Schickler 2008). These analyses challenge to Carmines and Stimson’s (1989) issue evolution perspective in two ways. First, the 88th Congress (1963-64, which Carmines and Stimson identify as the critical moment) is entirely unremarkable. No brief “burst of rapid change” occurs around this “flashpoint,” during which these authors argue that elites first take strong, clear and distinct positions on civil rights (Carmines and Stimson 1989, 154). In fact, our results do not point to any particular Congress or year as a critical moment. Rather, the parties distinguish themselves on civil rights incrementally. We observe a long, secular realignment – not a burst of rapid change at a critical moment followed by a longer period of dynamic growth, during which the

37

It is also worth noting that the results are robust to alternative specifications; in particular, we employed a zeroinflated negative binomial model to account for the overabundance of zero scores and obtained substantively similar results.

33

political system continues to slowly and asymptotically realign. Second, these results cast doubt on Carmines and Stimson’s claim that members of Congress and other national elites, as the first movers in an elites-activists-masses channel, have relatively broad discretion in determining their parties’ positions. The incremental nature of the partisan shift – and the breadth of that change across state parties and members of Congress -- suggests that the transformation was rooted in an earlier shift among the core constituencies of the Democratic coalition rather than the choices of particular elites.

Coalition Partners and Partisan Change What factors led nonsouthern Democrats to become the pro-civil rights delegation and Republicans to curb their support for civil rights? The fact that the Democrats supported civil rights by the early 1940s undercuts Carmines and Stimson’s explanation: that the actions of elites such as Lyndon Johnson and Barry Goldwater played the decisive role. Indeed, the incremental nature of this shift, over many decades, suggests that it is unlikely that the shift can be attributed to any one event or person. We argue that the Democratic Party’s core coalition partners were instrumental in leading the party towards its pro-civil rights position. These pro-civil rights coalition partners included not only civil rights organizations, but also labor unions, religious- and ethnic-affiliated groups (such as the Anti-Defamation League, American Jewish Committee, American Jewish Congress, and the Catholic Church) and broad-based progressive policy groups (most notably Americans for Democratic Action). These organizations, which were drawn into the Democratic Party during the New Deal for reasons having little or nothing to do with race, became among the most

34

vocal proponents of civil rights measures, pulling the party towards their position (Anderson 1964; Berman 1979; Kesselman 1948; Sitkoff 1971). Labor emerged as an important force early on. Farhang and Katznelson (2005) show that the Congress of Industrial Organizations’ strong support for civil rights contributed to the “braiding of labor and race politics.” The Democrat-affiliated United Auto Workers, Ladies Garment Workers Union and the Amalgamated Clothing Workers of America also lent their institutional voices to the civil rights cause (Anderson 1964). Even the traditionally more conservative AF of L, with CIO prodding, supported fair employment practices legislation in 1945 (Kesselman 1948, 151). African American voters, who flocked to the Democratic Party as early as 1936, also encouraged nonsouthern Democrats to embrace civil rights liberalism. In addition, organizations associated with ethnic or religious groups lobbied the Democratic Party on civil rights. The AntiDefamation League, American Jewish Committee, American Jewish Congress, and the Catholic Church all played key roles in both the Democratic and civil rights coalitions (Anderson 1964; Berman 1979, Chapter 2). With African Americans, Jews, and other ethnic minorities ensconced in the Democratic coalition, popular perceptions of the parties shifted so that, by the 1950s, the Democrats were seen as the party of minority groups and the Republicans as the party of the Protestant establishment (Berman 1979; Brock 1962). Progressive policy groups, most prominently Americans for Democratic Action (ADA), reinforced the connections between Democratic politics and civil rights advocacy. Indeed, 110 of the 1,234 delegates to the 1948 Democratic convention were ADA members (Brock 1962). ADA organized a series of rallies and public events to lobby other Democrats on civil rights during the 1948 presidential campaign (Sitkoff 1971, 606). Individuals such as Walter Reuther

35

and Joseph Rauh -- with one foot planted in the Democratic Party leadership and the other in these pro-civil rights groups -- encouraged the party’s progress on civil rights (Sindler 1962, 233). While Feinstein and Schickler (2008) examine how the Democrats’ coalitional partners influenced the civil rights planks in Democratic national and state-level platforms throughout the 1940s and 1950s, the connections between these groups and Democrats in Congress were similarly strong. When the Union for Democratic Action (the predecessor group to the ADA), the ACLU, the NAACP, and various Christian and Jewish groups joined together in 1943 to form the National Council for a Permanent Fair Employment Practices Commission, congressional northern Democrats, who drew many of their activist supporters from these groups, had good reason to promote the Council’s agenda (Kesselman 1948, 29-31). As early as 1944, for instance, Time magazine claimed that nonsouthern Democrats and Republicans supported an anti-poll tax measure because of “prodd[ing] by church, liberal, labor and Negro organizations.”38 Since the Democratic Party was more closely linked with these groups, the Democrats were more likely to be targeted by, and sympathetic to, such efforts than Republicans. When in 1952-53 Walter Reuther (CIO president, former UAW president, and a founder of ADA) and others called upon the Senate to end its rule permitting unlimited debate, which had doomed many civil rights bills, conservative Republicans and southern Democrats banded together to oppose the proposal (Fite 2002, 329). Nonsouthern Democrats supported the move, including presidential candidate Adlai Stevenson, whom Senator Richard Russell (D-GA) bemoaned was “influenced strongly by the CIO Political Action Committee and the Americans for Democratic Action (Fite 2002, 297).”

38

Time, 22 May 1944; as quoted in Bolling 1965, 19.

36

Later that decade, unions and the NAACP together played a key role in the passage of the Civil Rights Act of 1957 (Bolling 1965, 175-176). According to Rep. Richard Bolling (D-MO), although this bill did not go as far as many civil rights advocates had hoped, it was the most progressive bill that had a reasonable chance of passage. In order for it to pass, Bolling believes, these civil rights movement leaders had to refrain from publicly criticizing it, and there were only “four persons … [who] could play a role in persuading civil rights groups to refrain from undiluted public disparagement” of the bill: Andrew Biemiller, Kenneth Birkhead, Senator Paul Douglas (D-IL), and Joseph Rauh, Jr. (Bolling 1965, 175-176). These individuals each held a mix of both Democratic and coalition group attachments. Biemiller was the chief congressional lobbyists for the AFL-CIO and a former Democratic member of Congress from Wisconsin; Birkhead was the national director of the American Veterans’ Committee and former holder of a number of important positions in the Democratic Party;39 Douglas was active in the ADA (Brock 1962); and Rauh headed the Civil Rights Coalition, served as ADA chairman, was a member of numerous Democratic national platform committees, worked as general counsel of the UAW, and sat on the NAACP board of directors. These individuals, whom Bolling asserts were vital to “selling” the 1957 Act to civil rights activists beyond Capitol Hill, all served as key linkages between the Democratic Party and its interest groups allies that advocated civil rights legislation. The Republican Party, by contrast, did not experience similar pressure from its key constituencies. Instead, the GOP’s connections with the Chamber of Commerce, manufacturers’ groups, real estate organizations and other business interests pulled the party in the opposite direction on as early as the mid-1940s, on issues such as fair employment practices legislation (Chen 2006; 2007). Kesselman’s (1948) study of early legislative efforts emphasizes

39

E.g., Assistant to the Democratic National Committee chairman, Democratic Senatorial Campaign Committee staff director; assistant to the Senate Majority Whip.

37

congressional Republicans’ resistance to job discrimination protections for African Americans. Kesselman points to several factors to explain the lack of Republican support for FEPC. These include the “reluctance of Republican congressional leaders to antagonize the southern Democratic bloc in Congress” given that they had been working with the GOP to thwart Roosevelt and Truman’s domestic programs. In addition, Kesselman cites a newspaper report of a House Republican caucus from 1945 that suggested that about 90% of party members opposed a permanent FEPC Act, primarily because of their more general opposition to government bureaucracy (1948, 201-02). Furthermore, Republicans increasingly doubted that the bill would help the party win over African American voters. Perhaps the most frank statement along those lines came from House Speaker Joe Martin’s (R-MA) comments before an African American audience in 1947: “The FEPC plank in the 1944 Republican platform was a bid for the Negro vote, and they did not accept the bid … I’ll be frank with you: we are not going to pass a FEPC bill, but it has nothing to do with the Negro vote. We are supported by New England and Middle Western industrialists who would stop their contributions if we passed a law that would compel them to stop religious as well as racial discrimination in employment.”40 Notwithstanding Martin’s speech, however, Republicans generally tried to avoid taking a direct public stand against civil rights for fear of further alienating the party’s remaining African American supporters, along with pro-civil rights northerners. Thus, Kesselman’s account of the 1945 FEPC bill concludes that Republican leaders hoped that the bill would never reach the House floor, “sparing many Republicans from having to vote against their convictions for fear of the reaction of minority voters back home” (203). The weak GOP support for civil rights discharge petitions from the early 1940s onwards is consistent with Kesselman’s account, though 40

Pittsburgh Courier, January 4, 1947, 4; quoted in Berman, The Politics of Civil Rights in the Truman Administration, 59.

38

we cannot be certain whether most congressional Republicans were sincere opponents of civil rights, indifferent to the issue, or simply attempting to maintain their informal coalitional alliance with southern Democrats by helping them keep civil rights off of the agenda. We plan to explore this latter issue in future work. Even the case of the Civil Rights Act of 1957 – which was a focus of Carmines and Stimson’s analysis – points to the limited Republican commitment to civil rights in this period. Attorney General Herbert Brownell was a genuine, ardent pro-civil rights Republican and he played a crucial entrepreneurial role in pushing a reluctant Eisenhower Administration to (tepidly) support his push for a bill with enforcement powers (Anderson 1964). Yet it is striking the extent to which Brownell had to maneuver against the majority of Republicans in the Administration and congressional GOP leaders in unsuccessfully seeking a strong bill. While southern Democrats are widely blamed for weakening the Civil Rights Act, Eisenhower and most GOP leaders in Congress evidently also favored only modest legislation with limited enforcement powers (see, e.g., Anderson 1964). For example, Senate Republican leader William Knowland (R-CA) fought to water down the bill before it was introduced (Anderson 1964; Montgomery and Johnson 1998: 213). Knowland’s biographers conclude that due to weakening amendments adopted en route to passage, “by the time the legislation was enacted into law, Knowland had come around to believing that perhaps passage of a civil rights bill was the right thing to do” (Montgomery and Johnson 1998, 213). A focus on roll call votes, however, misses the role of Knowland and other Republicans in supporting southern Democratic efforts to weaken civil rights legislation.

39

CONCLUSION Extending the analysis of congressional behavior on civil rights issues beyond analyses of roll call votes reveals that the conventional wisdom needs revision. In this paper, we explain why roll call voting analysis misses congressional Democrats’ early support for civil rights and Republicans’ drift away from their earlier civil rights advocacy. Using roll call votes as a measure of preferences misses other significant legislative behavior that, without the constraints of committee gatekeeping, reveals far more than the small number of bills that make it to the House floor during this era. Contra Carmines and Stimson’s issue evolution model, we find that the partisan shift in Congress on civil rights is not characterized by a critical juncture in 1963-64; instead, it was incremental and had taken hold long before Barry Goldwater’s challenge to Lyndon Johnson. Well before the early 1960s, northern Democrats had clearly embraced civil rights while Republicans had started to step away from their historic role as the leading advocates of protections for African Americans. Pressure from Democratic coalitional partners may account for the party’s gradual embrace of civil rights liberalism. Republicans, with a markedly different set of interest group allies, experienced no such pull. While the close links between key Democratic members of Congress and pro-civil rights group leaders provides some support for our coalition partners theory, we believe that an exploration of connections between Democratic members and interest group supporters in the electorate could further buttress our claims. A logical next step in this analysis, therefore, will be to use district-level data, where available, to attempt to explain which Democratic members became vigorous civil rights supporters early on, and also to explain

40

Republican divisions on the issue.41 We also believe that it may be possible to use both the binary information on who signed each petition and the data on the sequencing of signatures to develop a more refined estimate of member preferences on civil rights issues. Our study also has important implications for understanding the Democratic coalition in Congress that emerged from the New Deal. While Franklin Roosevelt may have been largely content to cater to the southern wing of his party by shying away from civil rights advocacy,42 most nonsouthern Democrats in Congress embraced an alternative strategy by the mid-1940s. Their early willingness to defy the southern Democrats suggests that the ultimate break-up of the New Deal coalition was built into the structure of the alignment that emerged during Roosevelt’s administration. The core supporters of the civil rights cause -- union leaders, African Americans, ADA liberals, and Jews – were the same groups which nonsouthern Democrats most depended upon for votes and activist support. Once grassroots civil rights advocates gained sufficient popular strength to force a point of decision, there was little doubt that the majority of Democrats who hailed from outside the south would be the strongest backers of civil rights legislation. The simultaneous willingness of nonsouthern congressional Republicans to provide tacit assistance to the southern Democrats in their efforts to stall civil rights legislation was also a harbinger of the future partisan alignment. Barry Goldwater’s embrace of racial conservatism thus represented the culmination of a long process of change within the Republican party, just as Lyndon Johnson’s transformation into a racial liberal reflected deep currents in the party he sought to lead.

41

Scott Adler has made available district level data, starting in 1943, that includes percent African American, percent blue-collar, percent foreign born, and several other potential covariates. Unfortunately, union density is evidently only available at the state-level. 42 See McMahon 2004, however, for an account emphasizing FDR’s contributions to the civil rights case.

41

REFERENCES Adler, E. Scott, and John Wilkerson, Congressional Bills Project: 1947-1972. Available on-line at www.congressionalbills.org [accessed 15 May 2008]. Asher, Herbert B. 1973. “The Learning of Legislative Norms.” The American Political Science Review, (67) 2: 499-513. Anderson, J.W. 1964. Eisenhower, Brownell, and the Congress. Tuscalossa: University of Alabama Press. Bensel, Richard F. 1984. Sectionalism and American Political Development: 1880-1980. Madison, Wisconsin: University of Wisconsin Press. Berman, William C. 1970. The Politics of Civil Rights in the Truman Administration. Columbus: Ohio State University Press. Beth, Richard S. 1990. “The Discharge Rule in the House: Procedure, History, and Statistics.” Congressional Research Service Report. Report 90-84. March. The Library of Congress. Beth, Richard S. 2003. “The Discharge Rule in the House: Recent Use in Historical Context.” Congressional Research Service Report for Congress. Report 97-856. April. The Library of Congress. Bolling, Richard. 1965. House Out of Order. New York: Dutton. Brady, David W., and Charles Bullock, III. 1980. “Is There a Conservative Coalition in the House?” The Journal of Politics, 42(2): 549-559 Brock, Clifton. 1962. Americans for Democratic Action: Its Role in National Politics. Washington: Public Affairs Press. Cannon, David. 1999. Race, Redistricting, and Representation: The Unintended Consequences of Black Majority Districts. Carmines, Edward G., and James A. Stimson. 1989. Issue Evolution: Race and the Transformation of American Politics. Princeton, NJ: Princeton University Press. Carmines, Edward G., and James Woods. 2002. “The Role of Party Activists in the Evolution of the Abortion Issue.” Political Behavior 24. Carmines, Edward, and Michael Wagner. 2006. “Political Issues and Party Alignments: Assessing the Issue Evolution Perspective,” Annual Review of Political Science 9. Caro, Robert. 2002. Master of the Senate. New York: Knopf.

42

Chen, Anthony. 2006. “‘The Hitlerian Rule of Quotas’: Racial Conservatism and the Politics of Fair Employment Legislation in New York State, 1941–1945,” Journal of American History 92: 1238. Chen, Anthony. 2007. “The Party of Lincoln and the Politics of State Fair Employment Practices Legislation in the North, 1945-64,” American Journal of Sociology 112: 1713–74. Civettini, Andrew and Eric Hines. 2005. “Misspecification Effects in Zero-Inflated Negative Binomial Regression Models: Common Cases.” Paper presented at the 2005 Annual Meeting of the Southern Political Science Association. Clapp, Charles. 1963. The Congressman: His Work as He Sees It. Washington: The Brookings Institution. Evans, Rowland, and Robert Novak. 1966. Lyndon B. Johnson: The Exercise of Power. New York: New American Library. Farhang, Sean, and Ira Katznelson. 2005. “The Southern Imposition: Congress and Labor in the New Deal and Fair Deal,” Studies in American Political Development 19. Feinstein, Brian D. and Eric Schickler. 2008. “Platforms and Partners: The Civil Rights Realignment Reconsidered.” Studies in American Political Development. Fite, Gilbert. 2002. Richard B. Russell, Jr., Senator from Georgia. Chapel Hill, NC: UNC Press. Hall, Richard. 1996. Participation in Congress. New Haven: Yale University Press. Clinton, Joshua, and John Lapinski. 2007. “The Relationship Between Laws and Roll Calls in the U.S. Congress, 1889-1994.” Working paper. Karol, David. “Realignment without Replacement: Issue Evolution and Ideological Change among Members of Congress” (paper presented at the Annual Meeting of the Midwest Political Science Association, Chicago, IL, April 15–18, 1999). Karol, David. Coalition Management: Explaining Party Position Change in American Politics. Forthcoming, Cambridge University Press. Katznelson, Ira, and John Lapinski. 2006. “The Substance of Representation: Studying Policy Content and Legislative Behavior,” in The Macropolitics of Congress. E. Scott Adler and John Lapinski eds. Princeton: Princeton University Press. Lapham, Lewis J. 1954. Party Leadership and the House Committee on Rules. Ph.D. Dissertation, Harvard University. Lee, Taeku. 2002. Mobilizing Public Opinion. Chicago: University of Chicago Press.

43

Kesselman, Louis. 1948. The Social Politics of FEPC: A Study in Reform Pressure Movements. Chapel Hill: University of North Carolina Press. Lubell, Samuel. 1952. The Future of American Politics. New York: Harper and Brothers. MacNeil, Neil. 1963. Forge of Democracy: The House of Representatives. New York: David McKay Inc. McMahon, Kevin J. 2004. Reconsidering Roosevelt on Race. Chicago: University of Chicago Press. Martin, John. 1979. Civil Rights and the Crisis of Liberalism: The Democratic Party, 1945-1976. Boulder, CO: Westview Press. Oleszek, Walter. 2004. Congressional Procedure and the Policy Process, 6th ed. Washington: CQ Press. Pearson, Kathryn and Eric Schickler. 2007. “Discharge Petitions, Agenda Control, and the Congressional Committee System 1927-1976.” Prepared for presentation at the History of Congress Conference at Princeton University, May 18-19, 2007. Poole, Keith T., and Howard Rosenthal. 1997. Congress: A Political-Economic History of Roll Call Voting. New York: Oxford University Press. Rae, Nicol C. 1994. Southern Democrats. New York and Oxford: Oxford University Press. Schickler, Eric. 2001. Disjointed Pluralism: Institutional Innovation and the Development of the U.S. House. Princeton: Princeton University Press. Schiller, Wendy. 2000. Partners and Rivals: Representation in the U.S. Senate. Princeton: Princeton University Press. Sindler, Allen P. 1962. “The Unsolid South: A Challenge to the Democratic Party,” in The Uses of Power: Seven Cases in American Politics, ed. Allen Sindler. New York: Harcourt, Brace, and World. Sitkoff, Harvard. 1971. “Harry Truman and the Election of 1948: The Coming of Age of Civil Rights in American Politics,” Journal of Southern History 37(4):597-616. Tomz, Michael, Jason Wittenberg, and Gary King. 2003. “CLARIFY: Software for Interpreting and Presenting Statistical Results,” Journal of Statistical Software, 8, no. 1. Wilson, Rick and Cheryl Young. 1997. “Cosponsorship in the U.S. Congress,” Legislative Studies Quarterly 22(1).

44